Yeah, we learned a lot from the experiments we did, but we weren’t sure at the end of it whether it was the right direction, and it was more clear that it would require a significantly larger investment to figure that out, so we had to pause it.
Other things we’re doing with AI may still lead back to search improvements in other ways. But we’ve had to set this down for now.
We aren’t ready to start forecasting on a time basis, but I think the monthly cadence for these updates should help us get a feel for it together over time.
At the moment, that linked pull request is probably the best thing available to learn more details. I think the folks working on it will share once things are a bit further along – they’ve just been pretty heads down getting things done.
Is there anything in particular you’re interested in learning more about this project?