Something amiss today (~530 EST) in the AI results from the address bar. In two independent successive browser-window queries, the initial response was character looping (an L followed by a period, followed by rows of exclamation points).
Update: Make that four times. I tried a new tab, and a new browser session.
More problems. Today (~745pm EST) it repeatedly provided false values even when presented with the fact I checked the cited URL.
The output continually asserted a company’s rating of 21.8/100, even though the site (that AI provided as support for its 21.8 value) said 84.6.
Unfortunately it’s a pretty common issue with Qwen 3 - the model we use for most of our traffic. We’ve raised an issue with the Qwen team and have made some config changes on our side but will probably have to wait until it gets solved. In the meantime, you should be able to regenerate the message with a different model whenever you run into it
Thanks for the update/details.