Array<\/b>
\nExamples of Google\u2019s AI Overviews going off the rails have quickly gone viral as the update has rolled out to the public, but while the company has had to disable AI on some particularly viral queries, Google reiterates that the \u201cvast majority\u201d of AI results are \u201chigh-quality.\u201d
Over the past couple of days, Google\u2019s AI Overviews have been caught spitting out incorrect, and sometimes potentially dangerous information. There are many examples of this floating around, with one particularly viral example telling people to put glue on pizza.
That\u2019s a potentially dangerous suggestion, but Google has stopped serving AI Overviews on the query that originally triggered it; \u201ccheese not sticking to pizza.\u201d Another dangerous query we\u2019ve noticed that AI Overviews are no longer appearing on is the temperature chicken must be cooked to. This doesn\u2019t seem to apply across the board, though, and it\u2019s worth noting that no two AI Overviews will be identical. Often, making the same search twice will result in the Overview re-wording the result or focusing on a different aspect of the answer.
Google, in a statement to 9to5Google, confirmed that it is \u201ctaking swift action where appropriate under our content policies\u201d in regards to some of these dangerous results.
Google further notes that many of the viral examples circulating right now are \u201cdoctored\u201d or couldn\u2019t be reproduced \u2013 it\u2019s a bit ironic that Google is pointing the finger at purposeful misinformation given that AI Overviews have been presenting other purposeful misinformation as fact \u2013 and said that the \u201cvast majority of AI Overviews provide high quality information.\u201d
A Google spokesperson explains:
The vast majority of AI Overviews provide high quality information, with links to dig deeper on the web. Many of the examples we\u2019ve seen have been uncommon queries, and we\u2019ve also seen examples that were doctored or that we couldn\u2019t reproduce. We conducted extensive testing before launching this new experience, and as with other features we\u2019ve launched in Search, we appreciate the feedback. We\u2019re 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.<\/p>\n