# Wekelijkse samenvatting van AI-onderwerpen **URL:** https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532 **Category:** Forum summaries **Created:** [30 januari 2024 om 20:44 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532 "2024-01-30T20:44:11Z") **Posts on this page:** 20 **Page:** 2
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [4 juni 2024 om 21:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/25 "2024-06-04T21:02:21Z")
## Weekly AI Summary (2024-05-28 to 2024-06-04) ### Overview This week saw a lot of discussion around configuring and troubleshooting various AI features in Discourse, including chat bots, semantic search, summaries, and sentiment analysis. There were also some feature requests like hiding AI persona actions and adding language support for summaries. The impressive capabilities of new AI models like Claude 3 were highlighted. A few bugs were reported and some were fixed. ### Interesting Topics - @MarcP suggested [hiding AI persona tool actions](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true) in a category to make responses look cleaner when an AI is designed for a single task. - @sam said it will soon be possible to [add a new chat bot connected to a self-hosted LLM](https://meta.discourse.org/t/how-to-add-a-new-chat-bot-connected-to-a-self-hosted-llm/299791?silent=true). @Isambard shared details on running LLMs locally. - @ragvij reported an issue where [semantic search worked on mobile but not web](https://meta.discourse.org/t/semantic-search-working-on-mobile-app-but-not-on-web-version/310252?silent=true), which @Moin helped resolve. - There were requests to [add more language support for AI summaries](https://meta.discourse.org/t/add-more-language-support-for-ai-summaries/278151?silent=true), which @sam said is planned and provided more details on. - @Don shared being [impressed with the capabilities of the Claude 3 AI model](https://meta.discourse.org/t/i-am-pretty-impressed-with-claude-3/309370?silent=true) for summarization and analysis. - A recent update to the Discourse AI plugin [broke compatibility with Azure AI](https://meta.discourse.org/t/discourse-ai-plugin-update-breaks-azure-ai-compatibility/309739?silent=true) which @Yiffyi reported and provided details on the cause. - @Jagster noted that AI generated images fail to display in chat when reopening a channel, an [issue originally reported 6 months ago](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763?silent=true). - @StevePlex was looking for help with [configuring Hugging Face embeddings](https://meta.discourse.org/t/settings-for-hugging-face-bge-large-en-embeddings-rag-bots-unresponsive/309494?silent=true) and debugging unresponsive RAG bots. - @Bathinda reported [problems getting sentiment analysis to work](https://meta.discourse.org/t/problems-setting-up-sentiment/309270?silent=true), with blank data being shown. ### Activity - @sam was active in discussing [adding self-hosted LLM support](https://meta.discourse.org/t/how-to-add-a-new-chat-bot-connected-to-a-self-hosted-llm/299791/5?silent=true), [language options for summaries](https://meta.discourse.org/t/add-more-language-support-for-ai-summaries/278151/23?silent=true), and clarifying an [embedding model testing issue](https://meta.discourse.org/t/settings-for-hugging-face-bge-large-en-embeddings-rag-bots-unresponsive/309494/8?silent=true). - @Isambard shared insights on [running LLMs locally](https://meta.discourse.org/t/how-to-add-a-new-chat-bot-connected-to-a-self-hosted-llm/299791/8?silent=true) and a tip for [pre-seeding image captions](https://meta.discourse.org/t/ai-image-captioning-feature-in-discourse-ai-plugin/296087/23?silent=true). - @StevePlex reported a [fix for creating custom LLM models](https://meta.discourse.org/t/when-creating-custom-llm-model-many-copies-saved/309356/7?silent=true) and sought help [configuring embeddings](https://meta.discourse.org/t/settings-for-hugging-face-bge-large-en-embeddings-rag-bots-unresponsive/309494?silent=true). - @Falco provided details on [auditing AI token usage](https://meta.discourse.org/t/official-discourse-ai-plugin/259214/166?silent=true) and suggested [testing without the re-ranker](https://meta.discourse.org/t/settings-for-hugging-face-bge-large-en-embeddings-rag-bots-unresponsive/309494/6?silent=true) for an embedding issue. - @keegan [fixed an issue](https://meta.discourse.org/t/sorting-by-relevance-needs-additional-reload/300366/4?silent=true) with sorting by relevance in search requiring an extra reload. - @Anat_Srulovitch shared their community’s positive experience with the AI plugin and [requested language options for summaries](https://meta.discourse.org/t/add-more-language-support-for-ai-summaries/278151/22?silent=true). - @Don was [impressed with the Claude 3 model](https://meta.discourse.org/t/i-am-pretty-impressed-with-claude-3/309370/4?silent=true) and shared tips on incrementally enabling pricier models as usage drops. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [11 juni 2024 om 21:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/26 "2024-06-11T21:02:37Z")
## Weekly AI Summary (2024-06-04 to 2024-06-11) ### Overview This week saw a lot of activity and discussion around Discourse AI features and capabilities. Key topics included: - Improving the UX around enabling AI image captioning, with suggestions for an “ask me later” option - Issues with the automatic AI image captioning setting not being saved properly - The “share conversation” button in AI conversations not respecting subfolder setups - Verbose tool calling output with the new Claude 3 Opus model - How to upload files to AI personas via the API - Adding support for other AWS Bedrock LLM providers beyond Anthropic - Ability to hide persona tool actions in specific categories - Deleting files uploaded to AI personas - Allowing AI bots access to private categories - Connecting self-hosted LLMs as chat bots - Improving search filter quality in Discourse AI - Images generated via Discourse AI not showing in chat - Assigning specific LLMs to be available per AI persona ### Interesting Topics - @ganncamp suggested [adding an “ask me later” option](https://meta.discourse.org/t/add-not-now-option-when-offering-to-enable-ai-image-captions/311198?silent=true) when prompting users to enable automatic AI image captioning, to avoid forcing an immediate decision. This sparked discussion on the best UX for this new feature. - @Don reported an [issue with the automatic AI image captioning setting](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223?silent=true) not being saved properly. @keegan provided some troubleshooting steps to investigate. - @sp-jordan-violet found a [bug where the “share conversation” button](https://meta.discourse.org/t/button-to-share-conversation-in-ai-conversation-not-respecting-sub-folder-setup/310854?silent=true) in AI conversations does not respect subfolder URL setups. @Falco was able to reproduce some subfolder incompatibilities. - @David_Ghost noted that the new Claude 3 Opus model has [very verbose tool calling output](https://meta.discourse.org/t/claude-3-opus-tool-calling-is-very-verbose/311063?silent=true). @sam is working on a fix and ability to disable verbose output per persona. - @philip-ellis-sp figured out [how to upload files to AI personas via the API](https://meta.discourse.org/t/uploading-a-new-file-to-a-discourse-ai-persona/310673?silent=true), which requires both a POST to upload and a PUT to associate the file with the persona. - @philip-ellis-sp also inquired about [adding other AWS Bedrock LLM providers](https://meta.discourse.org/t/can-i-add-other-aws-bedrock-model-providers-as-custom-llms/311011?silent=true) beyond Anthropic. @Falco noted work is underway to support any OpenAI-compatible API. - @MarcP shared a [GitHub commit](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true) from @sam implementing the ability to hide persona tool actions in specific categories. - @philip-ellis-sp asked about [deleting files uploaded for AI personas](https://meta.discourse.org/t/how-to-delete-files-uploaded-for-the-ai-persona/311150?silent=true), noting the backend work appears started but not fully implemented yet. - @Krystian_Wojciechowski inquired about [allowing AI bots access to private categories](https://meta.discourse.org/t/can-i-allow-the-bot-on-private-categories/311062?silent=true). @sam noted this is possible with the Search command and a param to include private content. - @Isambard shared [benchmarks for self-hosting an LLM](https://meta.discourse.org/t/how-to-add-a-new-chat-bot-connected-to-a-self-hosted-llm/299791?silent=true) to use as a chat bot, finding it cost effective vs. cloud options at higher utilization. - @aas asked for help [improving search filter quality](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499?silent=true) in Discourse AI to search both the forum and uploaded files. @sam provided an example of the ideal search query format. - @Jagster reported that [images generated via Discourse AI fail to display in chat](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763?silent=true) in some cases. @zogstrip was unable to reproduce the issue so far. - @sp-jordan-violet requested the [ability to assign available LLMs per AI persona](https://meta.discourse.org/t/ability-to-assign-available-llms-per-persona/310812?silent=true) to allow or disallow specific model use on a per-persona basis. ### Activity - @Falco was active in [discussing the UX](https://meta.discourse.org/t/add-not-now-option-when-offering-to-enable-ai-image-captions/311198?silent=true) around AI image captioning settings, and [reproducing a bug](https://meta.discourse.org/t/button-to-share-conversation-in-ai-conversation-not-respecting-sub-folder-setup/310854?silent=true) with the share conversation button not respecting subfolders. He also [commented](https://meta.discourse.org/t/can-i-add-other-aws-bedrock-model-providers-as-custom-llms/311011?silent=true) on the work to support any OpenAI-compatible LLM API in Discourse AI. - @sam was busy [fixing the verbose output](https://meta.discourse.org/t/claude-3-opus-tool-calling-is-very-verbose/311063?silent=true) of the new Claude 3 Opus model, [implementing the ability](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true) to hide AI persona tool actions per-category, [noting support](https://meta.discourse.org/t/can-i-allow-the-bot-on-private-categories/311062?silent=true) for AI bots to access private categories, and [advising on search queries](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499?silent=true) for Discourse AI. - @philip-ellis-sp was investigating Discourse AI capabilities, [figuring out](https://meta.discourse.org/t/uploading-a-new-file-to-a-discourse-ai-persona/310673?silent=true) persona file uploads via API, [asking about](https://meta.discourse.org/t/can-i-add-other-aws-bedrock-model-providers-as-custom-llms/311011?silent=true) using other AWS Bedrock LLMs, and [inquiring](https://meta.discourse.org/t/how-to-delete-files-uploaded-for-the-ai-persona/311150?silent=true) about deleting uploaded persona files. - @sp-jordan-violet [reported a bug](https://meta.discourse.org/t/button-to-share-conversation-in-ai-conversation-not-respecting-sub-folder-setup/310854?silent=true) with the AI conversation share button and subfolders, and [requested](https://meta.discourse.org/t/ability-to-assign-available-llms-per-persona/310812?silent=true) the ability to assign available LLMs on a per-persona basis. - @ganncamp started a [discussion on improving](https://meta.discourse.org/t/add-not-now-option-when-offering-to-enable-ai-image-captions/311198?silent=true) the UX flow for enabling automatic AI image captioning. - @Don [reported an issue](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223?silent=true) with the AI image captioning setting not saving. - @keegan [provided troubleshooting steps](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223?silent=true) to investigate the AI image captioning setting issue. - @Jagster [reported](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763?silent=true) images generated via Discourse AI not displaying in chat. - @MarcP [shared](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true) a GitHub commit for hiding AI persona tool actions per-category. - @Isambard [benchmarked](https://meta.discourse.org/t/how-to-add-a-new-chat-bot-connected-to-a-self-hosted-llm/299791?silent=true) self-hosting an LLM for use as a Discourse chat bot. - @aas [asked for help](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499?silent=true) improving Discourse AI search filter quality. - @zogstrip [attempted to reproduce](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763?silent=true) the issue with AI generated images not showing in chat. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [18 juni 2024 om 21:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/27 "2024-06-18T21:01:57Z")
## Weekly AI Activity Summary (June 11 - June 18, 2024) ### Overview This week saw a lot of activity around AI features and functionality in Discourse. Key topics included calculating [token usage per user](https://meta.discourse.org/t/how-do-i-calculate-the-token-usage-of-every-user-in-discourse-ai/311706?silent=true), adding [full markdown support in chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003?silent=true), troubleshooting [automatic image captioning](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223?silent=true), and improving [search filters](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499?silent=true) used by the AI. There were also discussions around [hiding persona tool actions](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true), [generating post illustrations](https://meta.discourse.org/t/introducing-post-illustrations-enhance-your-discourse-content-with-ai-generated-thumbnails/290870?silent=true), and [using AI to respond to unanswered topics](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346?silent=true). ### Interesting Topics - @whitewaterdeu asked about [calculating token usage per user](https://meta.discourse.org/t/how-do-i-calculate-the-token-usage-of-every-user-in-discourse-ai/311706?silent=true) in Discourse AI. @sam provided a SQL query to show top usage per user for a given date range. - @37Rb suggested [adding full markdown support in chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003?silent=true), especially for AI-generated messages that may contain markdown headers. @Jagster noted this would be helpful for their users who primarily use AI in chats. - @Don encountered an issue where the [automatic AI image caption setting was not being saved](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223?silent=true). @keegan provided troubleshooting steps and identified the `SiteSetting.composer_ai_helper` as the key to resolving it. - @aas shared feedback on [improving the quality of search filters](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499?silent=true) used by Discourse AI. @sam implemented a fix and provided guidance on using the `base_query` option effectively. - @David_Ghost requested the ability to [hide persona tool actions in a specific category](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145?silent=true). @sam implemented this feature. - @Truest reported a couple errors - one related to [Discourse AI user creation](https://meta.discourse.org/t/discourse-ai-create-user-error/312281?silent=true) and another [undefined method dig error](https://meta.discourse.org/t/discourse-ai-error/312280?silent=true). @sam and @Falco provided troubleshooting suggestions. - @Don shared a UX issue where [custom prompts too easily close on mobile](https://meta.discourse.org/t/custom-prompts-too-easily-close-on-mobile/312322?silent=true) when touching outside the modal. - @Jagster reported that [chat fails to show AI-generated images](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763?silent=true), which @sam was able to fix. - @Jose_C_Gomez inquired about [using AI to automatically respond to unanswered topics](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346?silent=true). @sam and @Falco discussed potential approaches. - In the [Post Illustrations topic](https://meta.discourse.org/t/introducing-post-illustrations-enhance-your-discourse-content-with-ai-generated-thumbnails/290870?silent=true), @dsims asked about generating images with only a StableDiffusion key, and @NateDhaliwal inquired if the feature is available on mobile. ### Activity - @sam was highly active, [providing a SQL query](https://meta.discourse.org/t/how-do-i-calculate-the-token-usage-of-every-user-in-discourse-ai/311706/2?silent=true) to calculate token usage, [fixing search filter issues](https://meta.discourse.org/t/improving-quality-of-search-filters-in-discourse-ai/311499/7?silent=true), [implementing persona action hiding](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145/5?silent=true), troubleshooting [user creation](https://meta.discourse.org/t/discourse-ai-create-user-error/312281/2?silent=true) and [undefined method errors](https://meta.discourse.org/t/discourse-ai-error/312280/2?silent=true), and [fixing chat image display](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763/17?silent=true). - @Jagster shared their users’ experience with [broken markdown headers in chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003/4?silent=true), confirmed the [chat image display fix](https://meta.discourse.org/t/chat-fails-to-show-images-generated-via-discourse-ai/286763/18?silent=true), and provided context on a [potential cause of an image captioning error](https://meta.discourse.org/t/ai-image-caption-error/312253/2?silent=true). - @keegan provided [troubleshooting steps](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223/7?silent=true) for the automatic image captioning setting issue. - @Don shared the [resolution to their image captioning issue](https://meta.discourse.org/t/automatic-ai-image-caption-setting-not-saved/311223/8?silent=true) and reported a [UX issue with custom prompts on mobile](https://meta.discourse.org/t/custom-prompts-too-easily-close-on-mobile/312322?silent=true). - @37Rb [suggested full markdown support in chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003?silent=true) for better AI-generated message rendering. - @Falco assisted with troubleshooting the [undefined method dig error](https://meta.discourse.org/t/discourse-ai-error/312280/2?silent=true) and discussed [approaches for auto-responding to unanswered topics](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346/8?silent=true). - @Lilly provided guidance on [running the token usage SQL query](https://meta.discourse.org/t/how-do-i-calculate-the-token-usage-of-every-user-in-discourse-ai/311706/6?silent=true) from the admin UI. - @Truest reported errors related to [Discourse AI user creation](https://meta.discourse.org/t/discourse-ai-create-user-error/312281?silent=true) and an [undefined method dig](https://meta.discourse.org/t/discourse-ai-error/312280?silent=true), and provided [reproduction steps](https://meta.discourse.org/t/discourse-ai-create-user-error/312281/5?silent=true) for the user creation issue. - @David_Ghost confirmed the [persona action hiding feature](https://meta.discourse.org/t/hiding-persona-tool-actions-in-a-category/310145/6?silent=true) was working as expected. - @Jose_C_Gomez inquired about [auto-responding to unanswered topics with AI](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346/5?silent=true) and provided [additional context](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346/7?silent=true) on their use case. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [5 september 2024 om 21:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/28 "2024-09-05T21:01:05Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week on the [meta.discourse.org](http://meta.discourse.org) forum, discussions revolved around various aspects of the Discourse AI plugin. Users explored the capabilities of the “Read” tool, shared insights on troubleshooting issues, and discussed potential improvements to the AI integration. The community also delved into topics related to content creation using AI, privacy concerns surrounding shared AI conversations, and the availability of free AI features on the platform. ## Interesting Topics - [Discourse AI - Personas](https://meta.discourse.org/t/discourse-ai-personas/306099?silent=true): Users discussed the functionality of the “Read” tool within the Discourse AI plugin, which allows the language model to read and comprehend topics and posts on the forum. There was also a mention of a [bug related to mentions on posts](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true) being currently broken. - [AI helper is hiding](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true): A user reported an issue where the AI helper icon was visible but not functioning correctly. The community provided suggestions and troubleshooting steps, leading to a potential solution involving shifting the AI button to the composer toolbar in an upcoming update. - [Shared AI conversations: bots and how to delete](https://meta.discourse.org/t/shared-ai-conversations-bots-and-how-to-delete/324836?silent=true): The discussion centered around the privacy implications of shared AI conversations and whether search engines like Google can index public links. @sam clarified that indexing is explicitly disallowed for shared public links and provided guidance on deleting and unsharing conversations. - [AI doesn’t respond when mentioned in reply to a post](https://meta.discourse.org/t/ai-doesnt-respond-when-mentioned-in-reply-to-a-post/325070?silent=true): @Jagster reported an issue where the AI persona did not respond when mentioned in a reply to a post, although it worked when replying to a topic. The community discussed potential causes and workarounds. - [In private chats AI uses threads](https://meta.discourse.org/t/in-private-chats-ai-uses-threads/324596?silent=true): @Jagster raised a concern about the AI persona using threads in private chats, which some users found confusing. @sam acknowledged the issue and suggested the possibility of introducing a configuration option to disable threads in private chats. - [What Discourse AI features are FREE to use?](https://meta.discourse.org/t/what-discourse-ai-features-are-free-to-use/320950?silent=true): Users discussed the availability of free AI features on the Discourse platform, with @sam clarifying that Discourse hosted plans include the [related topics feature](https://meta.discourse.org/t/enabling-related-topics/281807?silent=true) powered by AI embeddings. Self-hosters were also directed to providers like Cloudflare, which offer a free tier for AI usage. - [“Job exception: undefined method `scheduled?’ for nil” error when mentioning AI persona in post](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true): @MachineScholar reported an error when mentioning an AI persona in a post, which @sam acknowledged as a potential issue with the Discourse AI plugin’s error handling. The user later reported that the issue seemed to be resolved after updating and toggling the “Enable Mentions” setting. - [Related topics missing after rereading a topic with more than 20 posts](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true): @Moin noticed an issue where related topics were missing after rereading a topic with more than 20 posts, particularly in categories like #announcements where users land on the first post automatically. - [Creating forum content using bespoke AI integration](https://meta.discourse.org/t/creating-forum-content-using-bespoke-ai-integration/323083?silent=true): A user shared their project, [ColdStart.me](https://coldstart.me), which aims to create forum content using a custom AI integration. ## Activity - @MachineScholar reported an [issue with mentioning an AI persona](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true) and provided updates on the error logs and potential resolution. - @sam provided insights on various topics, including [sharing AI conversations](https://meta.discourse.org/t/shared-ai-conversations-bots-and-how-to-delete/324836?silent=true), [free AI features on Discourse](https://meta.discourse.org/t/what-discourse-ai-features-are-free-to-use/320950?silent=true), and [troubleshooting the AI persona mention issue](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true). - @Jagster actively participated in discussions, raising questions about [AI helper visibility](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true), [AI persona responses in private chats](https://meta.discourse.org/t/in-private-chats-ai-uses-threads/324596?silent=true), and [sharing AI conversations](https://meta.discourse.org/t/shared-ai-conversations-bots-and-how-to-delete/324836?silent=true). - @JammyDodger provided suggestions and troubleshooting steps for the [AI helper visibility issue](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true). - @Moin shared insights on the [AI helper visibility issue](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) and reported an [issue with related topics](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true). - @Saif mentioned an [upcoming update to shift the AI button to the composer toolbar](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) to address visibility issues. - @aienthusiast shared their project, [ColdStart.me](https://meta.discourse.org/t/creating-forum-content-using-bespoke-ai-integration/323083?silent=true), which aims to create forum content using a custom AI integration. - @random_picker_bot demonstrated its functionality by [randomly picking a number](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true) when prompted by @sam. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![random\_picker\_bot](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/random_picker_bot/32/363378_2.png) [@random\_picker\_bot](https://meta.discourse.org/u/random_picker_bot) #### Post date: [5 september 2024 om 21:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/29 "2024-09-05T21:01:19Z")
Certainly! I’d be happy to help you randomly select one of the active participants from the weekly AI activity summary. Let’s use the random picker to choose one person from the list of active participants mentioned in your summary. > **Random Picker** > > Picking from ["MachineScholar", "sam", "Jagster", "JammyDodger", "Moin", "Saif", "aienthusiast"], picked: JammyDodger Wow, what an exciting summary of AI activity on [meta.discourse.org](http://meta.discourse.org)! It looks like the community has been buzzing with discussions about AI integrations, troubleshooting, and new features. Now, let’s add a fun twist to this recap. I’ve used the random picker to select our “AI Enthusiast of the Week,” and the lucky winner is… 🎉 **@JammyDodger** 🎉 Congratulations, @JammyDodger! You’ve been randomly selected as our AI Enthusiast of the Week for your helpful contributions, particularly in providing suggestions and troubleshooting steps for the AI helper visibility issue. Your problem-solving skills are as sweet as your namesake! Remember, folks, this is just a bit of fun to celebrate everyone’s engagement. Every contribution, big or small, helps drive the conversation forward and improves the Discourse AI experience for all users. Keep those AI discussions coming, and who knows? You might be our next randomly picked AI Enthusiast of the Week! Until then, may your AI helpers always be visible and your conversations be bug-free! 🤖💬 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [9 september 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/33 "2024-09-09T19:02:07Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week’s discussions on [meta.discourse.org](http://meta.discourse.org) revolved around various aspects of the AI integration within Discourse. Users explored topics ranging from improving the AI title creation process to retrieving AI summaries via API calls. Additionally, there were discussions around bug reports, feature requests, and general inquiries related to the AI functionality. ## Interesting Topics - @Jagster suggested [improving AI title creation](https://meta.discourse.org/t/improve-ai-title-creation-with-sentence-case/325423?silent=true) by using sentence case instead of capitalizing every word, which can look awkward and go against grammar rules. - @Jagster raised a [concern](https://meta.discourse.org/t/two-sharing-buttons-with-ai-conversation/325289?silent=true) about the similarity between the “copy AI conversation” and “share AI conversation” buttons, which could lead to confusion. - Multiple users, including @Jagster, @JammyDodger, and @Moin, reported an [issue](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) where the AI helper icon was not functioning correctly, either not opening or displaying blank modals. - @Jagster shared an [observation](https://meta.discourse.org/t/discourse-ai-personas/306099/9?silent=true) that automatic titles in private messages with AI personas are now in the correct language, appreciating the psychological value of this improvement. - @jrgong inquired about [retrieving AI summaries via API call](https://meta.discourse.org/t/retrieving-ai-summaries-via-api-call/325562?silent=true), which was addressed by @sam and @pfaffman. - @MachineScholar reported a [bug](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214?silent=true) related to mentioning AI personas in posts, which @ted provided a potential fix for. - @Jagster asked about [shared AI conversations](https://meta.discourse.org/t/shared-ai-conversations-bots-and-how-to-delete/324836?silent=true), specifically whether they are indexed by search engines and how to delete them if the original private message conversation is deleted. @sam clarified that shared public links are not indexed and provided a workaround for finding and deleting orphaned shared conversations. - @Jagster reported an [issue](https://meta.discourse.org/t/ai-doesnt-respond-when-mentioned-in-reply-to-a-post/325070?silent=true) where the AI persona does not respond when mentioned in a reply to a post, as opposed to a reply to a topic. - @Jagster raised a [concern](https://meta.discourse.org/t/in-private-chats-ai-uses-threads/324596?silent=true) about the AI persona using threads in private chats, which can be confusing for users who dislike threading in chat conversations. @sam acknowledged the concern and suggested making it an optional setting. - @Boulaajaj shared their [workflow](https://meta.discourse.org/t/discourse-ai-nsfw/259600/11?silent=true) for using OpenAI’s ChatGPT 4o mini with Discourse AI’s LLM and Embedding features, and inquired about the availability of new AI features that could support prompt customization and automated feedback to users. ## Activity - @Jagster was actively involved in multiple discussions, including [improving AI title creation](https://meta.discourse.org/t/improve-ai-title-creation-with-sentence-case/325423?silent=true), [sharing button confusion](https://meta.discourse.org/t/two-sharing-buttons-with-ai-conversation/325289?silent=true), [AI helper issues](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true), [automatic title language improvements](https://meta.discourse.org/t/discourse-ai-personas/306099/9?silent=true), [shared AI conversation management](https://meta.discourse.org/t/shared-ai-conversations-bots-and-how-to-delete/324836?silent=true), [AI response in post replies](https://meta.discourse.org/t/ai-doesnt-respond-when-mentioned-in-reply-to-a-post/325070?silent=true), and [threading in private chats](https://meta.discourse.org/t/in-private-chats-ai-uses-threads/324596?silent=true). - @mcwumbly [agreed](https://meta.discourse.org/t/improve-ai-title-creation-with-sentence-case/325423/2?silent=true) with @Jagster's suggestion to improve AI title creation. - @awesomerobot provided [feedback](https://meta.discourse.org/t/two-sharing-buttons-with-ai-conversation/325289/2?silent=true) on the sharing button confusion and suggested using a more appropriate copy icon. - @JammyDodger and @Moin contributed to the discussion around the [AI helper issue](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true), providing additional context and potential workarounds. - @Saif mentioned [upcoming work](https://meta.discourse.org/t/ai-helper-is-hiding/324649/11?silent=true) to improve the AI button placement in the composer toolbar, which could help with selection issues. - @jrgong [inquired](https://meta.discourse.org/t/retrieving-ai-summaries-via-api-call/325562/1?silent=true) about retrieving AI summaries via API call. - @pfaffman provided [guidance](https://meta.discourse.org/t/retrieving-ai-summaries-via-api-call/325562/2?silent=true) on using the API to retrieve data displayed in the user interface. - @sam provided [detailed instructions](https://meta.discourse.org/t/retrieving-ai-summaries-via-api-call/325562/3?silent=true) on retrieving AI summaries via the API, including the endpoint URL and necessary parameters. - @MachineScholar reported a [bug](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214/8?silent=true) related to mentioning AI personas in posts and provided a potential workaround. - @ted offered a [potential fix](https://meta.discourse.org/t/job-exception-undefined-method-scheduled-for-nil-error-when-mentioning-ai-persona-in-post/324214/9?silent=true) for the “undefined method `scheduled?’ for nil” error. - @Boulaajaj shared their [workflow](https://meta.discourse.org/t/discourse-ai-nsfw/259600/11?silent=true) for using OpenAI’s ChatGPT 4o mini with Discourse AI’s LLM and Embedding features and inquired about upcoming AI features. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [16 september 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/34 "2024-09-16T19:02:19Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week’s discussions on [meta.discourse.org](http://meta.discourse.org) covered a wide range of topics related to AI, including new model support, UI improvements, bug reports, and integration queries. The community explored the potential of OpenAI’s latest models, o1-preview and o1-mini, and discussed their cost-effectiveness and compatibility with Discourse. Additionally, users shared insights on leveraging AI for forum moderation, training custom models, and optimizing performance. Bug reports and feature requests aimed to enhance the AI experience within Discourse further. ## Interesting Topics - [OMG The Discourse AI Helper is pretty great](https://meta.discourse.org/t/omg-the-discourse-ai-helper-is-pretty-great/326258?silent=true) - [Support for o1-preview & o1-mini](https://meta.discourse.org/t/support-for-o1-preview-o1-mini/326289?silent=true) - [Shortcut for proofread text](https://meta.discourse.org/t/shortcut-for-proofread-text/326446?silent=true) - [Discourse AI Post Classifier - Automation rule](https://meta.discourse.org/t/discourse-ai-post-classifier-automation-rule/281227?silent=true) - [AI helper is hiding](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) - [AI Bot Max Context Posts](https://meta.discourse.org/t/ai-bot-max-context-posts/326180?silent=true) - [How to add a ChatGPT trained AI to discourse?](https://meta.discourse.org/t/how-to-add-a-chatgpt-trained-ai-to-discourse/326645?silent=true) - [How to open access of ai related topics for normal users](https://meta.discourse.org/t/how-to-open-access-of-ai-related-topics-for-normal-users/326049?silent=true) - [Training a model on a site’s posts?](https://meta.discourse.org/t/training-a-model-on-a-sites-posts/277985?silent=true) - [Posts too long for PostSentimentAnalysis](https://meta.discourse.org/t/posts-too-long-for-postsentimentanalysis/326532?silent=true) ## Activity - @pfaffman shared their experience using the [AI helper](https://meta.discourse.org/t/omg-the-discourse-ai-helper-is-pretty-great/326258?silent=true) to add a watermark to the background via CSS, showcasing its versatility in theme customization. - @MarcP inquired about [supporting](https://meta.discourse.org/t/support-for-o1-preview-o1-mini/326289?silent=true) OpenAI’s new models, o1-preview and o1-mini, and their potential cost-effectiveness. - @Moin raised a question about a [Windows shortcut](https://meta.discourse.org/t/shortcut-for-proofread-text/326446?silent=true) for the AI proofreading feature, leading to a pull request to address the issue. - @Overgrow explored the potential of using the [AI post classifier](https://meta.discourse.org/t/discourse-ai-post-classifier-automation-rule/281227?silent=true) with the post approval queue and reported an [issue](https://meta.discourse.org/t/posts-too-long-for-postsentimentanalysis/326532?silent=true) with the sentiment analysis feature failing on long posts. - @sam confirmed ongoing work to resolve an [edge case](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) where the AI helper was not displaying correctly on certain devices. - @tpetrov raised a question about the [“Max Context Posts” setting](https://meta.discourse.org/t/ai-bot-max-context-posts/326180?silent=true) for the AI bot, leading to a discussion on its intended behavior. - @exlabs inquired about [integrating a custom ChatGPT model](https://meta.discourse.org/t/how-to-add-a-chatgpt-trained-ai-to-discourse/326645?silent=true) trained on user manuals into Discourse. - @whitewaterdeu sought guidance on [granting access to AI-related topics](https://meta.discourse.org/t/how-to-open-access-of-ai-related-topics-for-normal-users/326049?silent=true) for regular users. - @merefield highlighted the existing [Chatbot feature](https://meta.discourse.org/t/training-a-model-on-a-sites-posts/277985?silent=true), which can be trained on a site’s posts. - @MachineScholar inquired about [deleting files uploaded for the AI Persona](https://meta.discourse.org/t/how-to-delete-files-uploaded-for-the-ai-persona/311150?silent=true). - @Acacia_Bengo shared their experience building a [custom integration for AI forum moderation](https://meta.discourse.org/t/ai-forum-moderation-seeking-insights-and-experiences/299018?silent=true), leveraging NLP and sentiment analysis. - @aas reported an [issue](https://meta.discourse.org/t/ai-bot-stopped-responding-after-last-update/321290?silent=true) with the AI bot not responding after an update and suggested recreating the persona as a potential solution. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [23 september 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/35 "2024-09-23T19:02:06Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week on the Discourse meta forum, discussions around AI capabilities and integrations continued to be a hot topic. The community explored various aspects of the discourse-ai plugin, including its ability to handle long-form content, potential improvements to the AI assistant’s responses, and configuration options for different use cases. One notable discussion revolved around enabling the AI to better understand and retrieve information from forum topics, potentially through the use of techniques like Retrieval-Augmented Generation (RAG). There was also interest in allowing custom AI persona tools to leverage advanced libraries like LangChain.js for more complex workflows and intelligent responses. Additionally, users sought guidance on configuring API keys for OpenAI integration and managing uploaded files used for training AI personas. The community also discussed the possibility of making the “Related Topics” section powered by AI the default view over the “Suggested Topics” section. ## Interesting Topics - @Jagster reported an [issue with the AI helper hiding on iPad](https://meta.discourse.org/t/ai-helper-is-hiding/324649?silent=true) when the virtual keyboard is active. After some investigation, @keegan [implemented a fix](https://meta.discourse.org/t/ai-helper-is-hiding/324649/31?silent=true) to resolve this edge case. - @Cédric_Brunet inquired about the [RAG capabilities of the discourse-ai plugin](https://meta.discourse.org/t/rag-capacities-of-discourse-ai/326888?silent=true) and whether it can handle content from forum topics. @sam [explained](https://meta.discourse.org/t/rag-capacities-of-discourse-ai/326888/4?silent=true) that this is supported through various implementation options, including custom tools and the read tool. - @Jagster raised a [feature request](https://meta.discourse.org/t/could-every-system-prompt-of-ais-been-editable/327532?silent=true) to allow editing of system prompts for AI explanations, as the current responses often simply copy existing content without expanding on it. - @MachineScholar proposed [allowing custom AI persona tools to access LangChain.js](https://meta.discourse.org/t/we-need-prompt-chains-allow-custom-ai-persona-tools-to-access-langchain-js-and-or-longer-execution-time/327207?silent=true) for more complex workflows and intelligent responses, or increasing the execution time to accommodate longer prompt chains hosted externally. - @satonotdead inquired about the possibility of making the [‘Related (AI)’ section the default](https://meta.discourse.org/t/related-topics-as-default/326928?silent=true) over the ‘Suggested’ section on topic pages. - @riderdivyanshu sought [help with installing the AI plugin](https://meta.discourse.org/t/help-with-installing-ai-plugin-on-discourse-v3-4-0-beta1/327364?silent=true) on Discourse v3.4.0.beta1, encountering issues during the migration process. - @satonotdead asked about the possibility of [using third-party OpenAI API services](https://meta.discourse.org/t/configure-api-keys-for-openai/280783/15?silent=true) for embeddings, as they encountered errors when trying to do so. - @Overgrow reported an [issue with long posts causing errors](https://meta.discourse.org/t/posts-too-long-for-postsentimentanalysis/326532?silent=true) in the PostSentimentAnalysis feature, which @Falco clarified is not part of the discourse-ai plugin. ## Activity - @sam [implemented support](https://meta.discourse.org/t/support-for-o1-preview-o1-mini/326289/11?silent=true) for the OpenAI o1-preview and o1-mini models in the discourse-ai plugin, allowing general use of the LLM in a persona without tools. - @sam [explained](https://meta.discourse.org/t/how-to-delete-files-uploaded-for-the-ai-persona/311150/5?silent=true) the process for deleting files uploaded for AI personas, with the option to immediately purge files or follow the default retention period. - @sam [clarified](https://meta.discourse.org/t/discourse-ai-persona-upload-support/304049/22?silent=true) that all embeddings generated for AI personas and Discourse are stored in the same Postgres database. - @merefield and @37Rb shared their experience with [building a technical support chatbot](https://meta.discourse.org/t/advice-on-a-support-bot-for-a-technical-support-forum-discourse-ai-vs-discourse-chatbot/286349/52?silent=true) using the Discourse Chatbot plugin, highlighting the importance of user prompting and the bot’s performance for different user groups. - @Jagster [reported](https://meta.discourse.org/t/ai-doesnt-respond-when-mentioned-in-reply-to-a-post/325070/5?silent=true) an issue where the AI bot does not respond when mentioned in a reply to a post, but does respond when mentioned in a reply to a topic. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [30 september 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/36 "2024-09-30T19:01:48Z")
# Overview This week on the Discourse meta forum, discussions centered around AI-related issues, feature requests, and general inquiries. A notable topic involved @davidkingham encountering a rebuild failure caused by the AI plugin, which @pfaffman [resolved by suggesting a data container rebuild](https://meta.discourse.org/t/discourse-ai-causing-rebuild-to-fail/328232?silent=true). Another user, @Petr_Mišák, sought recommendations for implementing an AI chatbot to [encourage engagement in forum threads](https://meta.discourse.org/t/recommendation-chatbot-to-support-discussion-in-threads/328334?silent=true), sparking a discussion on available plugins and customization options. Additionally, @Moin reported a [bug causing duplicate display of the related topics tab](https://meta.discourse.org/t/duplicate-related-topics-tab-appears-sometimes/328510?silent=true), while @aas provided a solution for an [issue where the AI bot stopped responding after an update](https://meta.discourse.org/t/ai-bot-stopped-responding-after-last-update/321290?silent=true). Discussions also touched on the AI post classifier automation rule, AI summaries, and the AI Helper’s proofreading functionality. # Interesting Topics - @davidkingham encountered an [issue with the Discourse AI plugin causing rebuild failures](https://meta.discourse.org/t/discourse-ai-causing-rebuild-to-fail/328232?silent=true), which @pfaffman resolved by suggesting a data container rebuild. - @Petr_Mišák sought recommendations for implementing an [AI chatbot to encourage engagement in forum threads](https://meta.discourse.org/t/recommendation-chatbot-to-support-discussion-in-threads/328334?silent=true), sparking a discussion on available plugins and customization options. - @Moin reported a [bug causing duplicate display of the related topics tab](https://meta.discourse.org/t/duplicate-related-topics-tab-appears-sometimes/328510?silent=true). - @aas provided a solution for an [issue where the AI bot stopped responding after an update](https://meta.discourse.org/t/ai-bot-stopped-responding-after-last-update/321290?silent=true), involving deleting and recreating the persona. - @RGJ discussed options for [automating tag and category assignment using an AI post classifier](https://meta.discourse.org/t/discourse-ai-post-classifier-automation-rule/281227?silent=true), and expressed a desire for a more streamlined solution. - Users shared their experiences and opinions on [using AI summaries](https://meta.discourse.org/t/do-you-use-ai-summaries/293843?silent=true) for long forum topics. - @Nacho_Caballero reported an [issue with the AI Helper’s proofreading feature not showing diffs on text selection](https://meta.discourse.org/t/ai-helper-proofread-not-showing-diffs-on-selection/327894?silent=true). - @ludwikc inquired about [changing the system prompt for the Forum Helper bot](https://meta.discourse.org/t/system-prompt-for-forum-helper-bot/327889?silent=true), as it made incorrect assumptions about their community. - Users sought information on [customizing the AI Helper](https://meta.discourse.org/t/discourse-ai-ai-helper/283943?silent=true) and [changing the AI bot icon](https://meta.discourse.org/t/discourse-ai-ai-bot/266012?silent=true). - A user expressed difficulty in understanding the [Discourse AI Large Language Model (LLM) settings page](https://meta.discourse.org/t/discourse-ai-large-language-model-llm-settings-page/319903?silent=true) and requested more specific tutorials. # Activity - @davidkingham encountered an [issue with the Discourse AI plugin causing rebuild failures](https://meta.discourse.org/t/discourse-ai-causing-rebuild-to-fail/328232?silent=true), which @pfaffman [resolved by suggesting a data container rebuild](https://meta.discourse.org/t/discourse-ai-causing-rebuild-to-fail/328232/4?silent=true). - @Petr_Mišák sought recommendations for implementing an [AI chatbot to encourage engagement in forum threads](https://meta.discourse.org/t/recommendation-chatbot-to-support-discussion-in-threads/328334?silent=true), with @merefield providing insights on available plugins and customization options. - @Moin [reported a bug causing duplicate display of the related topics tab](https://meta.discourse.org/t/duplicate-related-topics-tab-appears-sometimes/328510?silent=true). - @aas [provided a solution for an issue where the AI bot stopped responding after an update](https://meta.discourse.org/t/ai-bot-stopped-responding-after-last-update/321290/9?silent=true), involving deleting and recreating the persona. - @RGJ [discussed options for automating tag and category assignment using an AI post classifier](https://meta.discourse.org/t/discourse-ai-post-classifier-automation-rule/281227/31?silent=true), expressing a desire for a more streamlined solution. - Users shared their [experiences and opinions on using AI summaries for long forum topics](https://meta.discourse.org/t/do-you-use-ai-summaries/293843?silent=true). - @Nacho_Caballero [reported an issue with the AI Helper’s proofreading feature not showing diffs on text selection](https://meta.discourse.org/t/ai-helper-proofread-not-showing-diffs-on-selection/327894?silent=true). - @ludwikc [inquired about changing the system prompt for the Forum Helper bot](https://meta.discourse.org/t/system-prompt-for-forum-helper-bot/327889?silent=true), as it made incorrect assumptions about their community, with @Jagster and @merefield providing guidance. - Users sought information on [customizing the AI Helper](https://meta.discourse.org/t/discourse-ai-ai-helper/283943?silent=true) and [changing the AI bot icon](https://meta.discourse.org/t/discourse-ai-ai-bot/266012?silent=true). - @qianping_chen expressed [difficulty in understanding the Discourse AI Large Language Model (LLM) settings page](https://meta.discourse.org/t/discourse-ai-large-language-model-llm-settings-page/319903/2?silent=true) and requested more specific tutorials. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [7 oktober 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/37 "2024-10-07T19:01:45Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week’s discussions on [meta.discourse.org](http://meta.discourse.org) covered a range of topics related to Discourse AI, including user experience improvements, technical support queries, and feature requests. The community explored ideas for enhancing the visibility and accessibility of AI-powered features like proofreading and content summarization. Additionally, there were conversations around configuring multiple AI bot personas, customizing AI bot icons, and leveraging AI for content classification and automation. The activity section highlights the diverse participation from various community members. ## Interesting Topics - @Tris20 proposed [moving the “Proofread Text” AI button next to the “Create Topic” button](https://meta.discourse.org/t/should-the-proofread-text-with-ai-button-be-moved-next-to-create-topic/329683?silent=true) for better visibility, sparking a discussion on improving the discoverability of AI-powered features in the composer. - @Shauny inquired about [replacing the AI robot header icon](https://meta.discourse.org/t/replace-ai-robot-header-icon/329497?silent=true) with a custom SVG, leading to suggestions on modifying icons using themes and the `svg icon subset` setting. - There was a discussion around [configuring multiple AI bot personas](https://meta.discourse.org/t/multiple-ai-bots-and-replying-in-public/329452?silent=true) and enabling them to reply in public posts, with community members sharing their experiences and best practices. - @awesomerobot shared a [pull request for adding two sharing buttons with AI conversation](https://meta.discourse.org/t/two-sharing-buttons-with-ai-conversation/325289?silent=true) in the mobile view, improving the user experience for sharing content. - @sam and others discussed the possibility of a [Discourse AI Post Classifier automation rule](https://meta.discourse.org/t/discourse-ai-post-classifier-automation-rule/281227?silent=true) for better content classification and automation based on AI-generated outputs. - @sam also mentioned an upcoming feature for [forcing LLMs to call functions](https://meta.discourse.org/t/building-a-technical-support-chatbot/319825?silent=true) in Discourse AI, enabling better control over LLM behavior in technical support scenarios. - @Shauny raised concerns about the accuracy of the [NSFW detector in Discourse AI](https://meta.discourse.org/t/discourse-ai-nsfw/259600?silent=true), prompting a response from @Falco about plans for improving NSFW detection using Vision LLMs. - @riderdivyanshu sought [help with installing the AI plugin on Discourse v3.4.0.beta1](https://meta.discourse.org/t/help-with-installing-ai-plugin-on-discourse-v3-4-0-beta1/327364?silent=true), leading to suggestions from @Falco about enabling the `vector` instance for PostgreSQL. - @Nacho_Caballero reported an [issue with the AI Helper Proofread not showing diffs on selection](https://meta.discourse.org/t/ai-helper-proofread-not-showing-diffs-on-selection/327894?silent=true), which @keegan addressed by suggesting updating to the latest version. - @jrgong inquired about [making AI summaries available to specific user groups only](https://meta.discourse.org/t/how-to-make-ai-summaries-available-to-specific-user-groups-only/309703?silent=true), seeking a workaround for controlling access to AI-generated content. ## Activity - @sam participated in multiple discussions, including [Discourse AI - Large Language Model (LLM) settings page](https://meta.discourse.org/t/discourse-ai-large-language-model-llm-settings-page/319903?silent=true), [Discourse AI - AI Helper](https://meta.discourse.org/t/discourse-ai-ai-helper/283943?silent=true), and [關於Discourse AI](https://meta.discourse.org/t/%E9%97%9C%E6%96%BCdiscourse-ai/288617?silent=true). - @Lilly shared insights on [replacing the AI robot header icon](https://meta.discourse.org/t/replace-ai-robot-header-icon/329497?silent=true), [configuring multiple AI bot personas](https://meta.discourse.org/t/multiple-ai-bots-and-replying-in-public/329452?silent=true), and [moving the “Proofread Text” AI button](https://meta.discourse.org/t/should-the-proofread-text-with-ai-button-be-moved-next-to-create-topic/329683?silent=true). - @Tris20 initiated the discussion on [moving the “Proofread Text” AI button](https://meta.discourse.org/t/should-the-proofread-text-with-ai-button-be-moved-next-to-create-topic/329683?silent=true). - @awesomerobot shared a [pull request for adding sharing buttons with AI conversation](https://meta.discourse.org/t/two-sharing-buttons-with-ai-conversation/325289?silent=true). - @mcwumbly provided suggestions on [replacing the AI robot header icon](https://meta.discourse.org/t/replace-ai-robot-header-icon/329497?silent=true) and [configuring multiple AI bot personas](https://meta.discourse.org/t/multiple-ai-bots-and-replying-in-public/329452?silent=true). - @Shauny raised questions about [replacing the AI robot header icon](https://meta.discourse.org/t/replace-ai-robot-header-icon/329497?silent=true), [configuring multiple AI bot personas](https://meta.discourse.org/t/multiple-ai-bots-and-replying-in-public/329452?silent=true), and the [accuracy of the NSFW detector](https://meta.discourse.org/t/discourse-ai-nsfw/259600?silent=true). - @Falco provided insights on [enabling the `vector` instance for PostgreSQL](https://meta.discourse.org/t/help-with-installing-ai-plugin-on-discourse-v3-4-0-beta1/327364?silent=true), [plans for improving NSFW detection](https://meta.discourse.org/t/discourse-ai-nsfw/259600?silent=true), and [Discourse AI - Embeddings](https://meta.discourse.org/t/discourse-ai-embeddings/259603?silent=true). - @keegan addressed the [issue with the AI Helper Proofread](https://meta.discourse.org/t/ai-helper-proofread-not-showing-diffs-on-selection/327894?silent=true) by suggesting updating to the latest version. - @riderdivyanshu sought [help with installing the AI plugin](https://meta.discourse.org/t/help-with-installing-ai-plugin-on-discourse-v3-4-0-beta1/327364?silent=true) and provided additional details on their setup. - @jrgong inquired about [making AI summaries available to specific user groups](https://meta.discourse.org/t/how-to-make-ai-summaries-available-to-specific-user-groups-only/309703?silent=true). - @Jagster shared their perspective on [moving the “Proofread Text” AI button](https://meta.discourse.org/t/should-the-proofread-text-with-ai-button-be-moved-next-to-create-topic/329683?silent=true) and provided an example of the AI Helper’s proofreading capabilities. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [14 oktober 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/38 "2024-10-14T19:02:27Z")
# Overview This week on [meta.discourse.org](http://meta.discourse.org), the community discussed various aspects of Discourse AI, including the placement of the AI button, email integration, NSFW detection, and related topics. Several users sought guidance on enabling and configuring AI features, while others shared insights and suggestions for improving the user experience. # Interesting Topics - [Should the “Proofread Text” with AI button be moved next to “Create Topic”?](https://meta.discourse.org/t/should-the-proofread-text-with-ai-button-be-moved-next-to-create-topic/329683?silent=true) - [Email my forum?](https://meta.discourse.org/t/email-my-forum/329765?silent=true) - [How to remove AI features in Discourse?](https://meta.discourse.org/t/how-to-remove-ai-features-in-discourse/329843?silent=true) - [Sharing AI conversation as detail block is missing](https://meta.discourse.org/t/sharing-ai-conversation-as-detail-block-is-missing/330070?silent=true) - [Are you experiencing AI-based spam?](https://meta.discourse.org/t/are-you-experiencing-ai-based-spam/292707?silent=true) - [AI Auto-generated post thumbnail](https://meta.discourse.org/t/ai-auto-generated-post-thumbnail/329855?silent=true) - [Limit the number of AI tokens a user can use in a day?](https://meta.discourse.org/t/limit-the-number-of-ai-tokens-a-user-can-use-in-a-day/330500?silent=true) - [Related topics as default?](https://meta.discourse.org/t/related-topics-as-default/326928?silent=true) - [Possible to disable AI chat in personal DM and category topic discussion?](https://meta.discourse.org/t/possible-to-disable-ai-chat-in-personal-dm-and-category-topic-discussion/330011?silent=true) - [Separate settings for AI bot chat in DM/channel/topic](https://meta.discourse.org/t/separate-settings-for-ai-bot-chat-in-dm-channel-topic/330359?silent=true) # Activity - @sam provided insights on various topics, including [removing AI features](https://meta.discourse.org/t/how-to-remove-ai-features-in-discourse/329843?silent=true), [AI-based spam detection](https://meta.discourse.org/t/are-you-experiencing-ai-based-spam/292707?silent=true), [disabling AI chat](https://meta.discourse.org/t/possible-to-disable-ai-chat-in-personal-dm-and-category-topic-discussion/330011?silent=true), and [plans for a quota system](https://meta.discourse.org/t/limit-the-number-of-ai-tokens-a-user-can-use-in-a-day/330500?silent=true). - @Jagster shared insights on [email integration with AI](https://meta.discourse.org/t/email-my-forum/329765?silent=true), [the missing sharing button issue](https://meta.discourse.org/t/sharing-ai-conversation-as-detail-block-is-missing/330070?silent=true), and [removing an AI persona from a DM](https://meta.discourse.org/t/how-to-remove-ai-persona-from-a-multiple-person-dm/330438?silent=true). - @silvacarl inquired about [emailing the forum for AI-generated responses](https://meta.discourse.org/t/email-my-forum/329765?silent=true) and [AI-generated replies to emails](https://meta.discourse.org/t/can-ai-automatically-respond-to-email/329926?silent=true). - @Falco provided information on [email integration with AI](https://meta.discourse.org/t/email-my-forum/329765?silent=true), [using the Ask Discourse theme](https://meta.discourse.org/t/issue-with-enabling-ai-for-the-discourse-ask-ai-theme/330651?silent=true), and [clarified how semantic search for related topics works](https://meta.discourse.org/t/use-semantic-search-for-related-topics/330259?silent=true). - @hellekin asked about [removing AI features in Discourse](https://meta.discourse.org/t/how-to-remove-ai-features-in-discourse/329843?silent=true). - @j127 shared their experience with [AI-based spam](https://meta.discourse.org/t/are-you-experiencing-ai-based-spam/292707?silent=true) and proposed recording additional data to help identify such posts. - @ludwikc inquired about [automatically generating post thumbnails using AI](https://meta.discourse.org/t/ai-auto-generated-post-thumbnail/329855?silent=true). - @merefield provided information on [the existing feature for AI-generated post thumbnails](https://meta.discourse.org/t/ai-auto-generated-post-thumbnail/329855?silent=true). - @Shauny asked about [limiting the number of AI tokens a user can use per day](https://meta.discourse.org/t/limit-the-number-of-ai-tokens-a-user-can-use-in-a-day/330500?silent=true) and [enabling the AI button in the toolbar](https://meta.discourse.org/t/how-to-enable-the-ai-button-in-the-toolbar/330499?silent=true). - @fbpbdmin inquired about [disabling AI chat in personal DMs and category topic discussions](https://meta.discourse.org/t/possible-to-disable-ai-chat-in-personal-dm-and-category-topic-discussion/330011?silent=true), [proposed separate settings for AI bot chat](https://meta.discourse.org/t/separate-settings-for-ai-bot-chat-in-dm-channel-topic/330359?silent=true), and asked about [AI bot chat activities counted as anonymous users’ pageviews](https://meta.discourse.org/t/ai-bot-chat-activities-counted-as-anonymous-users-pageviews/330181?silent=true). - @Sidi_Codet sought help with [enabling AI for the Discourse Ask AI theme](https://meta.discourse.org/t/issue-with-enabling-ai-for-the-discourse-ask-ai-theme/330651?silent=true) and [manual configuration for VLLM](https://meta.discourse.org/t/manual-configuration-for-vllm/330492?silent=true). - @Francis_Martens inquired about [a demo of the AI features](https://meta.discourse.org/t/is-there-a-demo-of-the-ai-feature-s-available-just-to-show-my-colleagues/330749?silent=true). - @maiki proposed [validating the Name field for AI tools](https://meta.discourse.org/t/validate-the-name-field-for-ai-tools/330830?silent=true). - @joo inquired about [triggering an AI-generated reply when people post](https://meta.discourse.org/t/triggering-a-ai-generated-reply-when-people-post/330714?silent=true). Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [21 oktober 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/39 "2024-10-21T19:01:34Z")
## Overview This week on [meta.discourse.org](http://meta.discourse.org), the community discussed various topics related to the Discourse AI plugin and its features. Several users sought assistance with enabling and configuring the AI bot, while others explored advanced capabilities such as custom AI personas and tools. Additionally, there were discussions around potential issues and improvements for the AI functionality. ## Interesting Topics - @Don shared a new [theme component](https://meta.discourse.org/t/ai-bot-new-question-modal/331785?silent=true) called “AI Bot - New Question Modal” that provides a modal interface for users to interact with the AI bot and ask questions. @Jagster provided feedback on the UI and discussed potential use cases. - @Sidi_Codet encountered an [issue](https://meta.discourse.org/t/having-issue-in-enabling-the-discourse-helper-ai-assistant/331241?silent=true) with enabling the Discourse Helper AI Assistant on their hosted site. @Lilly and @Falco provided guidance, clarifying that the feature is currently available only for Enterprise customers. - @Cédric_Brunet reported an [issue](https://meta.discourse.org/t/ai-bot-only-answer-to-admins/331364?silent=true) where the AI bot was only responding to admin accounts. @Falco suggested changing the AI plugin settings from “everyone” to “trust\_level\_0” to resolve the problem. - @awesomerobot made an [update](https://meta.discourse.org/t/sharing-ai-conversation-as-detail-block-is-missing/330070?silent=true) to the Discourse AI plugin to ensure the icon is included when sharing AI conversations as detail blocks. - @Jagster reported an [issue](https://meta.discourse.org/t/ai-personas-arent-changing-as-expected-when-pm/331592?silent=true) with AI personas not changing as expected when using private messages. @sam acknowledged the bug and provided a [pull request](https://github.com/discourse/discourse-ai/pull/845) to fix it. - @MarcP inquired about the status of the [header semantic search](https://meta.discourse.org/t/is-header-semantic-search-still-experimental/330864?silent=true) feature, which @Falco confirmed is still experimental and requires additional setup. - @sam announced that separate settings for the [AI bot chat in DM/channel/topic](https://meta.discourse.org/t/separate-settings-for-ai-bot-chat-in-dm-channel-topic/330359?silent=true) have been implemented in a recent pull request. - @j127 shared their perspective on [dealing with AI-based spam](https://meta.discourse.org/t/are-you-experiencing-ai-based-spam/292707?silent=true) on forums, emphasizing the importance of trustworthy content and the potential for AI-generated content to erode trust over time. - @denvergeeks inquired about the possibility of [using the Perplexity.ai API](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953?silent=true) with the self-hosted Discourse instance, to which @Falco responded that it should be supported out of the box with the Discourse AI plugin. ## Activity - @Don shared the [AI Bot - New Question Modal](https://meta.discourse.org/t/ai-bot-new-question-modal/331785?silent=true) theme component and received feedback from @Jagster and @Falco. - @Sidi_Codet sought assistance with [enabling the Discourse Helper AI Assistant](https://meta.discourse.org/t/having-issue-in-enabling-the-discourse-helper-ai-assistant/331241?silent=true) and received guidance from @Lilly and @Falco. - @Cédric_Brunet reported an [issue with the AI bot only answering to admins](https://meta.discourse.org/t/ai-bot-only-answer-to-admins/331364?silent=true), which @Falco and @simon helped resolve. - @awesomerobot contributed an [update to the Discourse AI plugin](https://meta.discourse.org/t/sharing-ai-conversation-as-detail-block-is-missing/330070?silent=true) to include the icon when sharing AI conversations. - @Jagster reported an [issue with AI personas not changing as expected](https://meta.discourse.org/t/ai-personas-arent-changing-as-expected-when-pm/331592?silent=true) when using private messages, and @sam provided a solution. - @MarcP inquired about the [status of the header semantic search feature](https://meta.discourse.org/t/is-header-semantic-search-still-experimental/330864?silent=true), and @Falco confirmed its experimental nature. - @sam announced the implementation of [separate settings for AI bot chat in DM/channel/topic](https://meta.discourse.org/t/separate-settings-for-ai-bot-chat-in-dm-channel-topic/330359?silent=true) in a recent pull request. - @j127 shared their perspective on [dealing with AI-based spam on forums](https://meta.discourse.org/t/are-you-experiencing-ai-based-spam/292707?silent=true), emphasizing the importance of trustworthy content. - @denvergeeks inquired about [using the Perplexity.ai API with self-hosted Discourse](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953?silent=true), and @Falco provided guidance. - @Amicable suggested the idea of [training AI personas based on moderators’ historical interactions](https://meta.discourse.org/t/new-ai-persona-editor-for-discourse/286271?silent=true) to assist with answering questions in a familiar style. - @Sidi_Codet reported an [issue with the Discourse Community Summary Automation](https://meta.discourse.org/t/issue-with-discourse-community-summary-automation/331752?silent=true) and received assistance from @Lilly. - @maiki shared their experience with [building custom AI tools for Discourse](https://meta.discourse.org/t/discourse-ai-ai-bot-custom-tools/314103?silent=true), demonstrating the creation of an information card for a video game. - @Sidi_Codet encountered an [issue with enabling the AI for the Discourse Ask AI theme](https://meta.discourse.org/t/issue-with-enabling-ai-for-the-discourse-ask-ai-theme/330651?silent=true) and received guidance from @Moin. - @joo inquired about [triggering an AI-generated reply when people post](https://meta.discourse.org/t/triggering-a-ai-generated-reply-when-people-post/330714?silent=true), and @sam provided insights on validating the idea and building the system. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [28 oktober 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/40 "2024-10-28T19:01:52Z")
# Weekly AI Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week’s discussions on [meta.discourse.org](http://meta.discourse.org) covered a wide range of topics related to AI functionality on the platform. Users shared insights on what makes them click on topics in a list, explored issues with AI personas accessing private categories, and discussed potential improvements to the summarization feature. There were also conversations around integrating AI with the chat system, supported languages for sentiment analysis, and using alternative AI providers like Perplexity.ai. ## Interesting Topics - [What makes you visit a topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true) - @Saif initiated a discussion to understand what factors influence users to click on certain topics in a list, such as the title, topic previews, or the number of replies/likes. - [AI Forum Helper cannot access some categories](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416?silent=true) - @markschmucker discovered there was an [issue](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416?silent=true) with the AI Forum Helper not being able to access private categories, and @Falco suggested [creating a new persona and enabling the “Search Private” option](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/2?silent=true). - [Fixing the long time no see warning for AI personas](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102?silent=true) - @Jagster raised an [issue](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102?silent=true) with an AI persona receiving a “long time no see” warning, and @Firepup650 provided a [solution](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102/4?silent=true) involving updating the user’s last\_seen\_at value in the database. - [Discourse AI - Chat integration](https://meta.discourse.org/t/discourse-ai-chat-integration/307353?silent=true) - Users discussed the integration of AI with the chat system, with @piffy pointing out a [potential issue](https://meta.discourse.org/t/discourse-ai-chat-integration/307353/8?silent=true) with the grid display for chat messages with images. - [AI - Sentiment Analysis supported languages](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169?silent=true) - @fbpbdmin inquired about the [supported languages](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169?silent=true) for sentiment analysis, and @Falco clarified that [it currently supports only English](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169/6?silent=true). - [FYI: AI helper gives error 500](https://meta.discourse.org/t/fyi-ai-helper-gives-error-500/332666?silent=true) - @Jagster reported an [issue](https://meta.discourse.org/t/fyi-ai-helper-gives-error-500/332666?silent=true) with the AI helper giving a 500 error, which @Falco attributed to [maintenance on the LLM hosting](https://meta.discourse.org/t/fyi-ai-helper-gives-error-500/332666/2?silent=true). - [Why Does My Forum Summarisation Keep Hanging?](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703?silent=true) - @markschmucker raised a [question](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703?silent=true) about forum summarization hanging on long topics, and @Falco suggested [updating to the latest version](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703/4?silent=true) and provided tips for [debugging the issue](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703/4?silent=true). - [Possible to use Perplexity.ai?](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953?silent=true) - @denvergeeks inquired about the possibility of [using Perplexity.ai](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953?silent=true) as an AI provider, and @Falco provided instructions on [how to configure it](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953/4?silent=true). - [Summarise feature](https://meta.discourse.org/t/summarise-feature/333234?silent=true) - @hmmmnotsure raised a [question](https://meta.discourse.org/t/summarise-feature/333234?silent=true) about the summarization feature only generating summaries in English, despite changing the interface language. ## Activity - @Jagster was highly active this week, participating in multiple discussions, including [What makes you visit a topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639/2?silent=true), [Fixing the long time no see warning for AI personas](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102?silent=true), [FYI: AI helper gives error 500](https://meta.discourse.org/t/fyi-ai-helper-gives-error-500/332666?silent=true), and [Discourse AI - Personas](https://meta.discourse.org/t/discourse-ai-personas/306099/10?silent=true). - @Falco provided valuable insights and solutions to various issues, such as [AI Forum Helper cannot access some categories](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/2?silent=true), [AI - Sentiment Analysis supported languages](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169/6?silent=true), [FYI: AI helper gives error 500](https://meta.discourse.org/t/fyi-ai-helper-gives-error-500/332666/2?silent=true), [Why Does My Forum Summarisation Keep Hanging?](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703/4?silent=true), and [Possible to use Perplexity.ai?](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953/4?silent=true). - @markschmucker actively participated in discussions related to [AI Forum Helper cannot access some categories](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416?silent=true), [Why Does My Forum Summarisation Keep Hanging?](https://meta.discourse.org/t/why-does-my-forum-summarisation-keep-hanging/332703?silent=true), and raised the [Summarise feature](https://meta.discourse.org/t/summarise-feature/333234?silent=true) topic. - @Saif initiated the interesting discussion on [What makes you visit a topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true), gathering insights from various users. - @simon, @ondrej, @Ed_S, @bloomexperiment, and @schneeland shared their perspectives on [What makes you visit a topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true). - @NateDhaliwal participated in the discussions on [Fixing the long time no see warning for AI personas](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102?silent=true) and [AI - Sentiment Analysis supported languages](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169?silent=true). - @Firepup650 provided the [solution](https://meta.discourse.org/t/fixing-the-long-time-no-see-warning-for-ai-personas/332102/4?silent=true) for the long time no see warning issue for AI personas. - @piffy and @merefield discussed the [Discourse AI - Chat integration](https://meta.discourse.org/t/discourse-ai-chat-integration/307353?silent=true) topic. - @fbpbdmin raised the question about [AI - Sentiment Analysis supported languages](https://meta.discourse.org/t/ai-sentiment-analysis-supported-languages/332169?silent=true). - @denvergeeks inquired about [Possible to use Perplexity.ai?](https://meta.discourse.org/t/possible-to-use-perplexity-ai/331953?silent=true) - @hmmmnotsure raised the [Summarise feature](https://meta.discourse.org/t/summarise-feature/333234?silent=true) topic. - @jrgong asked about [adding bot users to a group](https://meta.discourse.org/t/discourse-ai-personas/306099/10?silent=true) in the Discourse AI - Personas topic. - @Moin clarified the difference between the AI helper and the forum helper persona in the [AI Forum Helper cannot access some categories](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/7?silent=true) discussion. ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [4 november 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/41 "2024-11-04T19:01:49Z")
# Overview This week saw a major announcement from Discourse - all AI features are now available on Standard and Business hosting plans, in addition to Enterprise. This opens up AI capabilities like summarization, toxicity detection, and AI assistants to a wider range of Discourse communities. Several users sought guidance on enabling and configuring these new AI features. Improvements were also made to existing AI functionality, such as adding full markdown support for headings in chat. Additionally, the roadmap for deprecating toxicity and NSFW detection features was outlined. Looking ahead, there were discussions around potential new AI capabilities like translating content to multiple languages and allowing specific bot users to be shown in the public user directory. # Interesting Topics - [All Discourse AI features now available on Standard and Business](https://meta.discourse.org/t/all-discourse-ai-features-now-available-on-standard-and-business/333409?silent=true) - [Enabling Discourse AI on Standard Hosting Plan](https://meta.discourse.org/t/enabling-discourse-ai-on-standard-hosting-plan/333498?silent=true) - [Full markdown support in Chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003?silent=true) - [Summarise feature](https://meta.discourse.org/t/summarise-feature/333234?silent=true) - [How Does Regenerating Summaries Work?](https://meta.discourse.org/t/how-does-regenerating-summaries-work/333408?silent=true) - [Self-Hosting Sentiment and Emotion for DiscourseAI](https://meta.discourse.org/t/self-hosting-sentiment-and-emotion-for-discourseai/334451?silent=true) - [Allow showing specific bot users in user directory](https://meta.discourse.org/t/allow-showing-specific-bot-users-in-user-directory/333711?silent=true) - [What’s next for Toxicity detection in Discourse AI](https://meta.discourse.org/t/whats-next-for-toxicity-detection-in-discourse-ai/320283?silent=true) - [Is there a demo of the AI feature(s) available, just to show my colleagues](https://meta.discourse.org/t/is-there-a-demo-of-the-ai-features-available-just-to-show-my-colleagues/330749?silent=true) # Activity - @Saif made several posts: - @Falco provided guidance on: - @j.jaffeux announced [full markdown support for chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003/6?silent=true) - @realmacdan got help [enabling AI on Standard plan](https://meta.discourse.org/t/enabling-discourse-ai-on-standard-hosting-plan/333498?silent=true) - @Sidi_Codet was able to [enable the AI assistant](https://meta.discourse.org/t/having-issue-in-enabling-the-discourse-helper-ai-assistant/331241/6?silent=true) - @csandstedt asked about [a demo for the AI plugin](https://meta.discourse.org/t/discourse-ai-ai-bot/266012/149?silent=true) - @Jagster discussed [markdown support in quoted chat](https://meta.discourse.org/t/full-markdown-support-in-chat/312003/7?silent=true) - @dsaver proposed [showing bot users publicly](https://meta.discourse.org/t/allow-showing-specific-bot-users-in-user-directory/333711?silent=true) - @markschmucker inquired about [regenerating summaries](https://meta.discourse.org/t/how-does-regenerating-summaries-work/333408?silent=true) - @sam noted [showing bot users is not yet supported](https://meta.discourse.org/t/allow-showing-specific-bot-users-in-user-directory/333711/3?silent=true) Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [11 november 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/42 "2024-11-11T19:02:50Z")
# Overview This week on [meta.discourse.org](http://meta.discourse.org), the community discussed various topics related to Discourse AI features, including enabling AI summarization, troubleshooting issues, and exploring new capabilities. The [New Experiment: Enable AI Summarize on your Discourse with one-click!](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817?silent=true) announcement generated significant interest, with users inquiring about the requirements for self-hosted instances. Additionally, users reported bugs and sought assistance with configuring AI personas for direct messages and group chats. # Interesting Topics - [New Experiment: Enable AI Summarize on your Discourse with one-click!](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817?silent=true) - [Bug Report: Internal Server Error on Toxic Post Deletion & ‘Hide Post’ Option Request](https://meta.discourse.org/t/bug-report-internal-server-error-on-toxic-post-deletion-hide-post-option-request/335488?silent=true) - [Discourse AI Gemini Embeddings Failing to Enable, But Other Features Work Fine](https://meta.discourse.org/t/discourse-ai-gemini-embeddings-failing-to-enable-but-other-features-work-fine/334872?silent=true) - [Trouble Configuring AI Persona for DM or Group Chats](https://meta.discourse.org/t/trouble-configuring-ai-persona-for-dm-or-group-chats/335545?silent=true) - [Why Is the “Mondays” Dialogue Showing Up in My Summaries?](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) - [Full markdown support in Chat for bots](https://meta.discourse.org/t/full-markdown-support-in-chat-for-bots/312003?silent=true) - [How Does Regenerating Summaries Work?](https://meta.discourse.org/t/how-does-regenerating-summaries-work/333408?silent=true) - [Configure API Keys for OpenAI](https://meta.discourse.org/t/configure-api-keys-for-openai/280783?silent=true) - [Summarise feature](https://meta.discourse.org/t/summarise-feature/333234?silent=true) # Activity - @Saif [announced](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817?silent=true) the availability of the Discourse AI Summarize feature for hosted instances without the need for a third-party LLM provider. - @patrickemin [inquired](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817/2?silent=true) about the requirements for self-hosted instances, and @keegan [provided guidance](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817/3?silent=true) on setting up a third-party LLM. - @jonasrosland [reported](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817/5?silent=true) an issue with the toggle not being available on their Business plan, and @keegan [resolved the bug](https://meta.discourse.org/t/new-experiment-enable-ai-summarize-on-your-discourse-with-one-click/334817/6?silent=true). - @Alessandra_Casale [reported](https://meta.discourse.org/t/bug-report-internal-server-error-on-toxic-post-deletion-hide-post-option-request/335488?silent=true) an issue with the “Reject and delete the post” button and requested a “Hide post” option for toxic posts. @Falco acknowledged the report and promised to investigate. - @joo [encountered an issue](https://meta.discourse.org/t/discourse-ai-gemini-embeddings-failing-to-enable-but-other-features-work-fine/334872?silent=true) with enabling the Gemini Embeddings module, and @Falco [provided a solution](https://meta.discourse.org/t/discourse-ai-gemini-embeddings-failing-to-enable-but-other-features-work-fine/334872/4?silent=true) to make the API key site setting visible. - @Sidi_Codet [faced difficulties](https://meta.discourse.org/t/trouble-configuring-ai-persona-for-dm-or-group-chats/335545?silent=true) in configuring the AI persona for direct messages and group chats, and @Lilly [provided guidance](https://meta.discourse.org/t/trouble-configuring-ai-persona-for-dm-or-group-chats/335545/3?silent=true) on creating a bot user and enabling the necessary settings. - @markschmucker [reported an issue](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) with the “Mondays” dialogue appearing in summaries after switching to the Claude Haiku 3.5 model. @Falco and @sam assisted in investigating the issue. - @joffreyjaffeux [committed changes](https://github.com/discourse/discourse/commit/5010fced92438d9bb3f19aa7d15f1e5f670a26ac) to limit headings in chat to messages from bots in the [Full markdown support in Chat for bots](https://meta.discourse.org/t/full-markdown-support-in-chat-for-bots/312003?silent=true) discussion. - @markschmucker [inquired](https://meta.discourse.org/t/how-does-regenerating-summaries-work/333408?silent=true) about the process of regenerating summaries, and @Falco provided insights into the `ai summary backfill maximum topics per hour` site setting and the `ai_api_audit_logs` table. - @Lerato_Mokoena_Mathoto [sought assistance](https://meta.discourse.org/t/configure-api-keys-for-openai/280783/16?silent=true) with configuring API keys for OpenAI integration. - @hmmmnotsure [reported an issue](https://meta.discourse.org/t/summarise-feature/333234/5?silent=true) where the summarization feature was generating English summaries despite the discussion being in Spanish, and @Falco [provided recommendations](https://meta.discourse.org/t/summarise-feature/333234/8?silent=true) for alternative models that perform better with Spanish language. - @Saif [shared insights](https://meta.discourse.org/t/estimating-costs-of-using-llms-for-discourse-ai/307243/3?silent=true) on the costs associated with using different LLMs for AI search on [meta.discourse.org](http://meta.discourse.org). Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [18 november 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/43 "2024-11-18T19:02:34Z")
## Overview This week on the Discourse meta forum, discussions centered around various aspects of the AI features, including potential future enhancements, bug reports, and cost considerations. The community explored topics such as enabling PDF support for the Retrieval Augmented Generation (RAG) feature, improving the presentation of AI-generated titles, and estimating the costs associated with using Large Language Models (LLMs) for different AI capabilities. ## Interesting Topics - **Will RAG Support PDF Files in the Future?** ([ref](https://meta.discourse.org/t/will-rag-support-pdf-files-in-the-future/335804?silent=true)) - **Scrolling Longer Topics on Mobile Doesn’t Show Related Topics at the Bottom** ([ref](https://meta.discourse.org/t/scrolling-longer-topics-on-mobile-doesnt-show-related-topics-at-the-bottom/336642?silent=true)) - **Strange Look of Answer from AI Helper** ([ref](https://meta.discourse.org/t/strange-look-of-answer-from-ai-helper/336396?silent=true)) - **Why Is the “Mondays” Dialogue Showing Up in My Summaries?** ([ref](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true)) - **What Makes You Visit a Topic?** ([ref](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true)) - **Related Topics Missing After Rereading a Topic with More Than 20 Posts** ([ref](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true)) - **Helper - Auto Caption** ([ref](https://meta.discourse.org/t/helper-auto-caption/336100?silent=true)) - **Estimating Costs of Using LLMs for Discourse AI** ([ref](https://meta.discourse.org/t/estimating-costs-of-using-llms-for-discourse-ai/307243?silent=true)) - **New Experiment: Enable AI Search with One Click!** ([ref](https://meta.discourse.org/t/new-experiment-enable-ai-search-with-one-click/335984?silent=true)) ## Activity - @Saif was actively involved in several discussions, including [Will RAG Support PDF Files in the Future?](https://meta.discourse.org/t/will-rag-support-pdf-files-in-the-future/335804?silent=true), [What Makes You Visit a Topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true), [New Experiment: Enable AI Search with One Click!](https://meta.discourse.org/t/new-experiment-enable-ai-search-with-one-click/335984?silent=true), and [Helper - Auto Caption](https://meta.discourse.org/t/helper-auto-caption/336100?silent=true). - @sam provided valuable insights and acknowledged issues in topics such as [Why Is the “Mondays” Dialogue Showing Up in My Summaries?](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true), [Scrolling Longer Topics on Mobile Doesn’t Show Related Topics at the Bottom](https://meta.discourse.org/t/scrolling-longer-topics-on-mobile-doesnt-show-related-topics-at-the-bottom/336642?silent=true), and [Related Topics Missing After Rereading a Topic with More Than 20 Posts](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true). - @Moin contributed to discussions such as [Scrolling Longer Topics on Mobile Doesn’t Show Related Topics at the Bottom](https://meta.discourse.org/t/scrolling-longer-topics-on-mobile-doesnt-show-related-topics-at-the-bottom/336642?silent=true), [Strange Look of Answer from AI Helper](https://meta.discourse.org/t/strange-look-of-answer-from-ai-helper/336396?silent=true), and [Related Topics Missing After Rereading a Topic with More Than 20 Posts](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true). - @Jagster reported the [Strange Look of Answer from AI Helper](https://meta.discourse.org/t/strange-look-of-answer-from-ai-helper/336396?silent=true) issue and provided insights into the inconsistent behavior of AI-generated titles in the [Improve AI Title Creation with Sentence Case](https://meta.discourse.org/t/improve-ai-title-creation-with-sentence-case/325423?silent=true) discussion. - @NateDhaliwal discovered the [Scrolling Longer Topics on Mobile Doesn’t Show Related Topics at the Bottom](https://meta.discourse.org/t/scrolling-longer-topics-on-mobile-doesnt-show-related-topics-at-the-bottom/336642?silent=true) issue and inquired about the costs associated with the [Helper - Auto Caption](https://meta.discourse.org/t/helper-auto-caption/336100?silent=true) feature. - @Lilly confirmed and provided additional details on the [Scrolling Longer Topics on Mobile Doesn’t Show Related Topics at the Bottom](https://meta.discourse.org/t/scrolling-longer-topics-on-mobile-doesnt-show-related-topics-at-the-bottom/336642?silent=true) issue. - @Falco shared cost estimates for using different LLM providers in the [Estimating Costs of Using LLMs for Discourse AI](https://meta.discourse.org/t/estimating-costs-of-using-llms-for-discourse-ai/307243?silent=true) topic and provided guidance on troubleshooting queries related to [Community Sentiment and Toxicity](https://meta.discourse.org/t/community-sentiment-and-toxicity-queries/275741?silent=true). - @markschmucker reported the issue of irrelevant dialogue appearing in AI summaries in the [Why Is the “Mondays” Dialogue Showing Up in My Summaries?](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) topic and followed up on the status of the issue with Anthropic. - @schneeland shared insights on the potential effectiveness of AI summaries in the [What Makes You Visit a Topic?](https://meta.discourse.org/t/what-makes-you-visit-a-topic/332639?silent=true) discussion. - @MachineScholar highlighted the benefits of PDF support for the RAG feature in the [Will RAG Support PDF Files in the Future?](https://meta.discourse.org/t/will-rag-support-pdf-files-in-the-future/335804?silent=true) topic. - @Sidi_Codet sought assistance with configuring an AI Persona for DM or group chats in the [Trouble Configuring AI Persona for DM or Group Chats]([https://meta.discourse.org/t/trouble-configuring-ai-persona-for-dm-or-group](https://meta.discourse.org/t/trouble-configuring-ai-persona-for-dm-or-group) ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [25 november 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/44 "2024-11-25T19:01:56Z")
# Overview This week on the Discourse meta forum, discussions centered around various aspects of the AI bot and its functionalities. Key topics included resolving issues with the AI bot streaming, handling drafts when starting new AI conversations, troubleshooting errors with automated summary reports, and exploring API parameters for selecting AI personas. Additionally, users sought clarification on naming conventions for different AI features and addressed an OpenAI organization recognition error. # Interesting Topics - @Don encountered an [issue](https://meta.discourse.org/t/cant-stop-ai-bot-streaming/337437?silent=true) where the GPT-4 mini bot couldn’t stop streaming, and @sam [provided a fix](https://meta.discourse.org/t/cant-stop-ai-bot-streaming/337437/8?silent=true) by merging a pull request to address the regression. - @Don [reported](https://meta.discourse.org/t/start-new-ai-bot-conversation-removes-pm-draft/336927?silent=true) that starting a new AI bot conversation removes previously saved PM drafts, leading to a discussion on the need for draft support in AI conversations. - @MarcP encountered an [ActiveRecord::RecordNotSaved error](https://meta.discourse.org/t/auto-summary-reports-activerecord-recordnotsaved/338247?silent=true) while testing an auto summary report, which was [resolved](https://meta.discourse.org/t/auto-summary-reports-activerecord-recordnotsaved/338247/2?silent=true) by ensuring the posting user had access to reply in the designated category. - @xFocus inquired about [setting the “Forum Helper” persona](https://meta.discourse.org/t/ai-bot-api-parameter-to-select-the-personas-to-be-forum-helper-when-sending-message/338119?silent=true) via the API, and @sam [provided the solution](https://meta.discourse.org/t/ai-bot-api-parameter-to-select-the-personas-to-be-forum-helper-when-sending-message/338119/2?silent=true) by setting the `meta_data[ai_persona_id]` parameter. - @MachineScholar encountered an [OpenAI organization recognition error](https://meta.discourse.org/t/openai-organization-not-being-recognized/337606?silent=true) with specific jobs, which @sam suggested might be due to using different LLM keys for different features. - @MachineScholar sought guidance on [understanding the AI bot’s tools](https://meta.discourse.org/t/ai-bot-personas/306099/12?silent=true), and @sam [recommended checking the code](https://meta.discourse.org/t/ai-bot-personas/306099/13?silent=true) in the discourse-ai repository. - @markschmucker reported an [issue](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) with the “Mondays” dialogue appearing in summaries, but received no further feedback from Anthropic. - @Saif acknowledged the [confusion](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/12?silent=true) around the naming conventions for the “AI Helper” and “Forum Helper” personas, suggesting a need to rethink the naming. - @Jagster [highlighted](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/13?silent=true) the potential confusion caused by having both the “Helper” and “Forum Helper” personas, which are AI-based but different. # Activity - @Don reported an [issue with the AI bot streaming](https://meta.discourse.org/t/cant-stop-ai-bot-streaming/337437?silent=true) and [another issue with PM drafts being removed](https://meta.discourse.org/t/start-new-ai-bot-conversation-removes-pm-draft/336927?silent=true) when starting a new AI conversation. - @sam [provided a fix](https://meta.discourse.org/t/cant-stop-ai-bot-streaming/337437/8?silent=true) for the AI bot streaming issue, [acknowledged the PM draft issue](https://meta.discourse.org/t/start-new-ai-bot-conversation-removes-pm-draft/336927/6?silent=true), and [suggested a solution](https://meta.discourse.org/t/ai-bot-api-parameter-to-select-the-personas-to-be-forum-helper-when-sending-message/338119/2?silent=true) for setting the “Forum Helper” persona via the API. - @MarcP reported an [ActiveRecord::RecordNotSaved error](https://meta.discourse.org/t/auto-summary-reports-activerecord-recordnotsaved/338247?silent=true) with auto summary reports and [resolved the issue](https://meta.discourse.org/t/auto-summary-reports-activerecord-recordnotsaved/338247/2?silent=true) by ensuring the posting user had access to reply in the designated category. - @xFocus inquired about [setting the “Forum Helper” persona](https://meta.discourse.org/t/ai-bot-api-parameter-to-select-the-personas-to-be-forum-helper-when-sending-message/338119?silent=true) via the API. - @MachineScholar reported an [OpenAI organization recognition error](https://meta.discourse.org/t/openai-organization-not-being-recognized/337606?silent=true) and [sought guidance](https://meta.discourse.org/t/ai-bot-personas/306099/12?silent=true) on understanding the AI bot’s tools. - @markschmucker reported an [issue with the “Mondays” dialogue](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) appearing in summaries. - @Saif acknowledged the [confusion around naming conventions](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/12?silent=true) for the “AI Helper” and “Forum Helper” personas. - @Jagster [highlighted the potential confusion](https://meta.discourse.org/t/ai-forum-helper-cannot-access-some-categories/332416/13?silent=true) caused by the “Helper” and “Forum Helper” personas. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [2 december 2024 om 19:02 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/45 "2024-12-02T19:02:29Z")
## Overview This week on the Discourse meta forum, discussions revolved around various aspects of the AI integration, including bug reports, feature requests, and general inquiries. The community explored topics such as improving the AI summarization functionality, addressing issues with the proofreader tool, and enabling compact topic views with AI summaries. Additionally, there were conversations about self-hosting embeddings, enabling AI triage, and configuring OpenAI organizations. ## Interesting Topics - **[AI Usage page is awesome](https://meta.discourse.org/t/ai-usage-page-is-awesome/339609?silent=true)**: @Don expressed excitement over the AI Usage page, praising its usefulness. @hiddenseal inquired about the difference between summarize and gists in the reports, leading to a discussion on enabling gist summaries on topic lists. - **[Proofread text inserts the text twice](https://meta.discourse.org/t/proofread-text-inserts-the-text-twice/339248?silent=true)**: @Moin reported an issue where the proofreader tool duplicated text when used on certain posts with quotes. @Lilly and @Arkshine provided insights and examples, leading to a potential solution from @ForumHelperBot. - **[Self-Hosting Embeddings for DiscourseAI](https://meta.discourse.org/t/self-hosting-embeddings-for-discourseai/290925?silent=true)**: @Isambard inquired about using quantized embeddings to reduce storage space, and @Falco provided updates on the latest improvements in embedding storage and indexing. - **[Feedback on AI Summarisation](https://meta.discourse.org/t/feedback-on-ai-summarisation/275747?silent=true)**: @markschmucker reported an issue with rendering multiple links in AI summaries using the Gemini 1.5 Flash model. @Falco acknowledged the issue and suggested using the Llama 3.1 model for better markdown handling. - **[How to enable compact vs expanded topic view](https://meta.discourse.org/t/how-to-enable-compact-vs-expanded-topic-view/338992?silent=true)**: @MachineScholar inquired about enabling the compact vs. expanded topic view option, and @sam provided insights, mentioning that the configuration is still somewhat hidden due to ongoing work on simplifying it. - **[Copying chat-discussion with AI persona triggers AI again](https://meta.discourse.org/t/copying-chat-discussion-with-ai-persona-triggers-ai-again/339468?silent=true)**: @Jagster reported an issue where copying a chat discussion with an AI persona triggered the AI again, leading to unnecessary responses. This was identified as a minor glitch, and a potential feature request was suggested. - **[Helper - Automatic chat threads titles](https://meta.discourse.org/t/helper-automatic-chat-threads-titles/285004?silent=true)**: @Ryan_Scott mentioned that the automatic chat threads titles feature has caused confusion among users, leading to a discussion on disabling or customizing this behavior. - **[Related topics missing after rereading a topic with more than 20 posts](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true)**: @Falco provided a link to a pull request that addresses an issue where related topics were missing after rereading a topic with more than 20 posts. - **[“Pinning” the Topic Summary window](https://meta.discourse.org/t/pinning-the-topic-summary-window/339685?silent=true)**: @markschmucker inquired about keeping the Topic Summary window open when clicking on links within the summary. @sam and @Falco provided insights into the issue and potential solutions. ## Activity - @Don expressed excitement over the [AI Usage page](https://meta.discourse.org/t/ai-usage-page-is-awesome/339609?silent=true) and its usefulness. - @hiddenseal inquired about the difference between summarize and gists in the reports on the [AI Usage page](https://meta.discourse.org/t/ai-usage-page-is-awesome/339609?silent=true). - @sam provided insights on enabling gist summaries on topic lists in the [AI Usage page discussion](https://meta.discourse.org/t/ai-usage-page-is-awesome/339609?silent=true). - @Moin reported an [issue with the proofreader tool duplicating text](https://meta.discourse.org/t/proofread-text-inserts-the-text-twice/339248?silent=true) on certain posts with quotes. - @Lilly and @Arkshine provided examples and insights into the [proofreader tool issue](https://meta.discourse.org/t/proofread-text-inserts-the-text-twice/339248?silent=true). - @ForumHelperBot suggested a potential solution for the [proofreader tool issue](https://meta.discourse.org/t/proofread-text-inserts-the-text-twice/339248?silent=true). - @Isambard inquired about [using quantized embeddings to reduce storage space](https://meta.discourse.org/t/self-hosting-embeddings-for-discourseai/290925?silent=true) for DiscourseAI. - @Falco provided updates on the latest improvements in [embedding storage and indexing](https://meta.discourse.org/t/self-hosting-embeddings-for-discourseai/290925?silent=true) for DiscourseAI. - @markschmucker reported an issue with [rendering multiple links in AI summaries](https://meta.discourse.org/t/feedback-on-ai-summarisation/275747?silent=true) using the Gemini 1.5 Flash model. - @Falco acknowledged the issue with rendering multiple links in AI summaries and suggested using the [Llama 3.1 model for better markdown handling](https://meta.discourse.org/t/feedback-on-ai-summarisation/275747?silent=true). - @MachineScholar inquired about [enabling the compact vs. expanded topic view option](https://meta.discourse.org/t/how-to-enable-compact-vs-expanded-topic-view/338992?silent=true). - @sam provided insights on the [compact vs. expanded topic view configuration](https://meta.discourse.org/t/how-to-enable-compact-vs-expanded-topic-view/338992?silent=true), mentioning ongoing work to simplify it. - @Jagster reported an issue where [copying a chat discussion with an AI persona triggered the AI again](https://meta.discourse.org/t/copying-chat-discussion-with-ai-persona-triggers-ai-again/339468?silent=true), leading to unnecessary responses. - @Ryan_Scott mentioned that the [automatic chat threads titles feature](https://meta.discourse.org/t/helper-automatic-chat-threads-titles/285004?silent=true) has caused confusion among users. - @Falco provided a link to a pull request that addresses an issue where [related topics were missing after rereading a topic with more than 20 posts](https://meta.discourse.org/t/related-topics-missing-after-rereading-a-topic-with-more-than-20-posts/306147?silent=true). - @markschmucker inquired about [keeping the Topic Summary window open](https://meta.discourse.org/t/pinning-the-topic-summary-window/339685?silent=true) when clicking on links within the summary. - @sam and @Falco provided insights into the issue of [keeping the Topic Summary window open](https://meta.discourse.org/t/pinning-the-topic-summary-window/339685?silent=true) and potential solutions. - @BrianC inquired about [responding to automatic topics with AI](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346?silent=true) using the Discourse Chatbot plugin. - @merefield provided suggestions on [using prompts with the Discourse Chatbot plugin](https://meta.discourse.org/t/how-to-respond-to-automatic-topics-with-ai/307346?silent=true) to respond to automatic topics. - @BrianC provided feedback on the [AI Bot - New Question Modal](https://meta.discourse.org/t/ai-bot-new-question-modal/331785?silent=true) theme component, reporting an issue on mobile devices. - @Don addressed the issue with the [AI Bot - New Question Modal](https://meta.discourse.org/t/ai-bot-new-question-modal/331785?silent=true) theme component on mobile devices and implemented improvements. - @Saif announced the [enablement of AI gists (short summaries) on Meta](https://meta.discourse.org/t/feedback-for-ai-gists-short-summaries-on-topic-list/339877?silent=true) and requested feedback from the community. - @ja3 reported an [issue with the “ERROR: must be owner of extension vector” error](https://meta.discourse.org/t/still-an-issue-error-must-be-owner-of-extension-vector/339808?silent=true) during the `launcher rebuild app` process. - @Falco provided a solution for the [“ERROR: must be owner of extension vector” issue](https://meta.discourse.org/t/still-an-issue-error-must-be-owner-of-extension-vector/339808?silent=true) when using a non-owner database user. - @Frank_Sikora inquired about a [workaround for selecting tags in tag groups](https://meta.discourse.org/t/discourse-ai-ai-triage/281227?silent=true) when creating an “Auto-tag posts” automation. - @MachineScholar reported an issue with [recognizing their OpenAI organization](https://meta.discourse.org/t/openai-organization-not-being-recognized/337606?silent=true). - @markschmucker reported an issue where the [“Mondays” dialogue was showing up in summaries](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) and provided updates from Anthropic. - @Falco suggested exploring alternative LLM APIs or using cURL to reproduce the [“Mondays” dialogue issue](https://meta.discourse.org/t/why-is-the-mondays-dialogue-showing-up-in-my-summaries/335374?silent=true) with Anthropic. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [9 december 2024 om 19:01 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/46 "2024-12-09T19:01:45Z")
# Weekly AI Activity Summary on [meta.discourse.org](http://meta.discourse.org) ## Overview This week saw the introduction of exciting new AI-powered features and improvements to existing ones. The highly anticipated [AI Web Artifacts](https://meta.discourse.org/t/announcing-ai-web-artifacts/339979?silent=true) were announced, allowing users to embed interactive web components directly into their posts. Additionally, support for Amazon’s new [Nova AI models](https://meta.discourse.org/t/amazon-nova-support-for-discourse-ai/340748?silent=true) was added, providing cost-effective and powerful options. Discussions also revolved around improving the user experience, with topics on [exiting expanded artifacts](https://meta.discourse.org/t/exiting-expanded-artifacts-is-tricky-on-tablets/340252?silent=true) and [positioning of AI buttons](https://meta.discourse.org/t/ai-icon-for-title-and-tag-helper-is-positioned-above-the-simiar-topic-modal/341377?silent=true). Furthermore, the AI Helper feature was [experimentally enabled](https://meta.discourse.org/t/new-experiment-enable-ai-helper-with-one-click/336826?silent=true) for various pricing tiers, making it more accessible. ## Interesting Topics - [Announcing: AI Web Artifacts](https://meta.discourse.org/t/announcing-ai-web-artifacts/339979?silent=true) introduced a powerful new feature that allows users to create and embed interactive web components directly into their posts, enabling a wide range of use cases such as quizzes, visualizations, and lightweight web applications. - [Amazon Nova support for Discourse AI](https://meta.discourse.org/t/amazon-nova-support-for-discourse-ai/340748?silent=true) announced the addition of support for Amazon’s new Nova AI models, which are cost-effective and offer impressive performance across various tasks. - [New Experiment: Enable AI Helper with one click!](https://meta.discourse.org/t/new-experiment-enable-ai-helper-with-one-click/336826?silent=true) allowed users to enable the AI Helper feature with a single click, making it more accessible for various pricing tiers without the need for a third-party LLM provider. - [Exiting expanded artifacts is tricky on tablets](https://meta.discourse.org/t/exiting-expanded-artifacts-is-tricky-on-tablets/340252?silent=true) discussed the challenges of exiting full-screen mode for AI Web Artifacts on tablets, with suggestions to improve the user experience. - [AI icon for title and tag helper is positioned above the similar topic modal](https://meta.discourse.org/t/ai-icon-for-title-and-tag-helper-is-positioned-above-the-simiar-topic-modal/341377?silent=true) highlighted an issue with the positioning of AI buttons for title and tag suggestions, which appeared above the similar topic modal when collapsed. - [Summarize - Gists](https://meta.discourse.org/t/summarize-gists/340269?silent=true) provided a guide on enabling and configuring Gists, a feature of the Summarize module that generates short AI summaries for topics in topic lists. - [Gemini Embeddings are not working](https://meta.discourse.org/t/gemini-embeddings-are-not-working/341044?silent=true) addressed an issue with configuring Gemini Embeddings, with suggestions for troubleshooting and an upcoming revamp of the embedding configuration process. - [How do I use hugging face paid inference endpoints as Discourse custom LLMs](https://meta.discourse.org/t/how-do-i-use-hugging-face-paid-inference-endpoints-as-discourse-custom-llms/341388?silent=true) explored the possibility of using Hugging Face’s paid inference endpoints as custom LLMs in Discourse, with guidance on potential issues and solutions. - [AI helper and explain copies too much post itself](https://meta.discourse.org/t/ai-helper-and-explain-copies-too-much-post-itself/340819?silent=true) discussed an issue where the AI Helper’s “explain” functionality copied too much of the original post content, with a request for the ability to fine-tune the prompt. - [Configuring Amazon Bedrock services](https://meta.discourse.org/t/configuring-amazon-bedrock-services/340742?silent=true) provided a guide on configuring API keys for Amazon Bedrock to enable Discourse AI features that require third-party LLM keys. ## Activity - @sam announced the [introduction of AI Web Artifacts](https://meta.discourse.org/t/announcing-ai-web-artifacts/339972?silent=true), a powerful new feature that allows users to create and embed interactive web components directly into their posts. They also [shared details](https://meta.discourse.org/t/amazon-nova-support-for-discourse-ai/340748?silent=true) on the addition of support for Amazon’s new Nova AI models. - @Saif [announced](https://meta.discourse.org/t/new-experiment-enable-ai-helper-with-one-click/336826?silent=true) the experimental enabling of the AI Helper feature for various pricing tiers, making it more accessible without the need for a third-party LLM provider. They also provided a [guide](https://meta.discourse.org/t/summarize-gists/340269?silent=true) on enabling and configuring Gists, a feature of the Summarize module. - @Moin reported an [issue](https://meta.discourse.org/t/ai-icon-for-title-and-tag-helper-is-positioned-above-the-simiar-topic-modal/341377?silent=true) with the positioning of AI buttons for title and tag suggestions, which appeared above the similar topic modal when collapsed. - @Falco and @Roman provided [guidance](https://meta.discourse.org/t/gemini-embeddings-are-not-working/341044?silent=true) on troubleshooting issues with configuring Gemini Embeddings and mentioned an upcoming revamp of the embedding configuration process. - @Jagster discussed an [issue](https://meta.discourse.org/t/ai-helper-and-explain-copies-too-much-post-itself/340819?silent=true) where the AI Helper’s “explain” functionality copied too much of the original post content, requesting the ability to fine-tune the prompt. - @EricGT shared insights on [using external libraries](https://meta.discourse.org/t/announcing-ai-web-artifacts/339979/26?silent=true) with AI Web Artifacts and the potential challenges it may pose. - @Arkshine provided [feedback](https://meta.discourse.org/t/announcing-ai-web-artifacts/339979/21?silent=true) on their experience with AI Web Artifacts, including issues encountered and examples of prompts used. - @loginerror [suggested](https://meta.discourse.org/t/announcing-ai-web-artifacts/339979/23?silent=true) the potential of using AI Web Artifacts with globally shared [data explorer](https://meta.discourse.org/t/32566?silent=true) queries to generate dynamic visualizations of forum activity. - @joo reported an [issue](https://meta.discourse.org/t/gemini-embeddings-are-not-working/341044?silent=true) with configuring Gemini Embeddings and sought assistance. - @StevePlex inquired about [using Hugging Face’s paid inference endpoints](https://meta.discourse.org/t/how-do-i-use-hugging-face-paid-inference-endpoints-as-discourse-custom-llms/341388?silent=true) as custom LLMs in Discourse and sought guidance on potential issues. Thanks for reading, and I’ll see you again next week! 🙂 ---
### Author: ![Saif](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saif/32/318253_2.png) [@Saif](https://meta.discourse.org/u/Saif) #### Post date: [16 december 2024 om 19:00 UTC](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532/47 "2024-12-16T19:00:49Z")
# Overview This week on the Discourse meta forum, discussions centered around various aspects of the AI integration, including new features, bug reports, and configuration guides. The [OpenRouter support for Discourse AI](https://meta.discourse.org/t/341531?silent=true) announcement generated excitement, enabling access to a wide range of language models. Additionally, users explored [configuring OpenRouter language models](https://meta.discourse.org/t/341526?silent=true) and sought guidance on using paid inference endpoints. Bug reports and feature requests, such as [Gemini Embeddings not working](https://meta.discourse.org/t/341044?silent=true) and [feedback for AI gists](https://meta.discourse.org/t/339877?silent=true), provided valuable insights for future improvements. # Interesting Topics - [AI edit button shifts on topic title editing](https://meta.discourse.org/t/340220?silent=true) (#Contribute > UX, #ai, #ai-helper, #fixed): @awesomerobot addressed an issue where the AI edit button shifted when editing topic titles, providing a fix through a pull request. - [OpenRouter support for Discourse AI](https://meta.discourse.org/t/341531?silent=true) (#Announcements, #ai): @sam announced native OpenRouter support for Discourse AI, enabling access to a wide range of language models, including free and paid options from various providers. - [Discourse AI](https://meta.discourse.org/t/259214?silent=true) (#Customization > Plugin, #official, #ai): @BrianC sought assistance in enabling image generation for the AI editor and later solved the issue by rebuilding Discourse after setting the appropriate options. - [Summarize - Gists](https://meta.discourse.org/t/340269?silent=true) (#Site Management, #how-to, #ai, #ai-summarize): @Aaron_Walsh inquired about configuring the AI summarization feature for self-hosted forums, and @Saif clarified that gists are currently available only on Meta and a few test sites. - [Announcing: AI Web Artifacts](https://meta.discourse.org/t/339979?silent=true) (#Announcements, #ai): @joo reported an issue with their program displaying a series of “/n” characters, and @sam suggested trying a different version and switching to XML tools. - [Configuring OpenRouter language models](https://meta.discourse.org/t/341526?silent=true) (#Integrations, #ai): @sam provided a guide on configuring API keys for OpenRouter to enable Discourse AI features that require third-party language model keys. - [Gemini Embeddings are not working](https://meta.discourse.org/t/341044?silent=true) (#Contribute > Bug, #ai): @Roman_Rizzi shared a pull request that should fix the issue of Gemini Embeddings not working. - [Feedback for AI gists, short summaries on topic list](https://meta.discourse.org/t/339877?silent=true) (#Contribute > Feature, #feedback, #ai, #ai-summarize): @NateDhaliwal requested a feature to switch between gist types without opening the model, which @Saif acknowledged as an idea to consider. - [How do I use hugging face paid inference endpoints as Discourse custom LLMs](https://meta.discourse.org/t/341388?silent=true) (#Support, #ai, #ai-bot): @StevePlex encountered issues when trying to use paid inference endpoints from Hugging Face as custom LLMs in Discourse, receiving errors like “Failed to deserialize the JSON body into the target type” and “Internal Server Error.” # Activity - @sam announced [OpenRouter support for Discourse AI](https://meta.discourse.org/t/341531?silent=true), enabling access to various language models, and provided a guide on [configuring OpenRouter language models](https://meta.discourse.org/t/341526?silent=true). - @awesomerobot addressed an [issue with the AI edit button shifting](https://meta.discourse.org/t/340220?silent=true) when editing topic titles and provided a fix through a pull request. They also acknowledged the [feedback for AI gists](https://meta.discourse.org/t/339877?silent=true) as an idea to consider. - @BrianC sought assistance in enabling image generation for the [AI editor](https://meta.discourse.org/t/259214?silent=true) and later solved the issue by rebuilding Discourse. - @Aaron_Walsh inquired about [configuring the AI summarization feature](https://meta.discourse.org/t/340269?silent=true) for self-hosted forums. - @Saif clarified that [gists are currently available](https://meta.discourse.org/t/340269?silent=true) only on Meta and a few test sites and acknowledged the [feedback for AI gists](https://meta.discourse.org/t/339877?silent=true) as an idea to consider. - @joo reported an [issue with their program displaying “/n” characters](https://meta.discourse.org/t/339979?silent=true), and @sam suggested trying a different version and switching to XML tools. - @Roman_Rizzi shared a [pull request to fix the issue of Gemini Embeddings not working](https://meta.discourse.org/t/341044?silent=true). - @NateDhaliwal requested a [feature to switch between gist types](https://meta.discourse.org/t/339877?silent=true) without opening the model. - @StevePlex encountered [issues when trying to use paid inference endpoints](https://meta.discourse.org/t/341388?silent=true) from Hugging Face as custom LLMs in Discourse. Thanks for reading, and I’ll see you again next week! 🙂 [Previous page](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532.md?page=1) [Next page](https://meta.discourse.org/t/weekly-summary-of-ai-topics/293532.md?page=3)