Google Is Taking ‘Swift Action’ To Remove Inaccurate AI Overview Responses

Earlier this month Google started rolling out its AI Overviews to all US users, a move that almost instantly produced reports of some pretty unusual responses. Now Google has responded by saying it is taking “swift action” to remove responses from the bot that are providing users with inaccurate information.

From saying the best way to keep toppings on pizza is with a little glue, to claiming President Andrew Jackson graduated from UW-Madison in 2005, accuracy doesn’t seem to be AI Overview’s strong suit.

That lack of accuracy might have a bit to do with where it’s getting its information. Google signed a licensing deal with Reddit in February, suggesting some of its responses will be generated from there. In fact, that glue-on-pizza tip can be attributed to a post by Reddit user f*cksmith 11 years ago.

This Tweet is currently unavailable. It might be loading or has been removed.

While Google has acknowledged that some of the answers AI Overviews produces are inaccurate, it claims that the inaccurate answers are few and far between.

“The vast majority of AI Overviews provide high-quality information, with links to dig deeper on the web. Many of the examples we’ve seen have been uncommon queries, and we’ve also seen examples that were doctored or that we couldn’t reproduce,” a Google representative told Business Insider. 

The representative went on to say that Google conducted extensive testing before launching the feature and that “we’re taking swift action where appropriate under our content policies, and using these examples to develop broader improvements to our systems, some of which have already started to roll out.”

That doesn’t mean you still aren’t likely to come across an incorrect answer or two. If you do get an AI Overview answer that seems a little questionable, you may want to do your own research before believing everything you read.