That’s great feedback, as we are just starting a rewrite of the UI for this feature.
6 Likes
In case it helps, for my community it would also be very useful to:
-
Be able to generate transcriptions for uploaded mp3 files.
-
Be able to fine-tune a model and use that for proofreading instead of the default.
1 Like
A simple solution could be showing a textarea with the proofread version next to (or below) the combined diff.
That way it would be possible to revert changes (since they’re shown in context) and make additional in-context modifications (punctuation, better word choices, etc.) before going back to the regular composer.
Suggestions noted in the Discourse AI topic that might fit with the Composer AI Helper
Have the AI suggest markdown edits.
(ref)
Have the AI read the first post of a topic to make sure
- It is coherent
- The category and the tags agree. (I know this can be configured but why not let the AI do it)
- The content of the post agrees with the category and tags.
- The content of the post agrees with the title of the post.
- It is relevant to the forum. (I know spam detectors will catch some of this but sometimes it is not spam and just doesn’t belong)
(ref)
1 Like
Suggestion
Note: This is for images with mostly text.
If a post contains an image, especially new users, then include an options to use OCR to extract the text from the image and show the text below the image.
Users who post just images, especially for source code, make it harder if the source code needs to be copied into a development environment.
Also images AFAIK are not converted to text before with the text being indexed by search engines. As such these post are less likely to be found via a search.
1 Like
Suggestion
Note: This is for images with no text.
For sites that use lots of images, add a option to create words or descriptions associated with the image and include in the post.
Example Discourse site that uses lots of images.
Note: Please add to the list or suggest other similar Discourse sites; personally interested to learn of other such sites.
AI for image classification
For an impressive use of image to text classification
https://valentin.deschaintre.fr/text2fabric_browser_v1.html
The related paper
“The Visual Language of Fabrics” by Valentin Deschaintre, Julia Guerrero-Viu, Diego Gutierrez, Tamy Boubekeur, Belen Masia (pdf)
2 Likes
Another suggestion: Change the language of the output. Currently, it defaults to English, but in our case we want the output to be in German.
1 Like
Hello. Will this plugin conflict if it is installed together with this chatbot plugin?
Can I, for example, connect GPT 3.5 TURBO in this plugin, and connect GPT4 in the chatbot plugin. Will this action cause a plugin conflict?
I have no idea, but if you give it a try let us know.
3 Likes
The guide has been updated to help you set up API keys for Discourse AI.
2 Likes
Good evening,
I’m using this Composer AI to help in our forum. However, it doesn’t appear as it’s shown in the Description above, where it adds the Composer AI Helper module button within the toolbar. as you can see below this is how it is
The issue is that the AI icon is behind the toolbar when I highlight the word, causing a problem when using a mobile device or a tablet.
when i hold down and swipe away i can see it
As you can see we’re changing the UI for this feature from a toolbar to a hover button on select.
It’s still underwork, but it should work on mobile already, at least it’s working on my phone
2 Likes
I’ve been testing both plugins for a week now and haven’t found a single conflict between them. For my version of the forum, each plugin does its job well 
It’s as if they were created to complement each other 
1 Like
Composer AI doesn’t work for me in Safari on Mac OS. I don’t get menu options after highlighting text and clicking the AI tool. It works fine in Chrome.
1 Like
Thanks @fidelio for making me aware of this. I can reproduce this issue and I will take a look at resolving it very soon!
3 Likes
This should now be resolved as of