Google’s electronic brain behind AI Overviews, a suspected audience killer for news publishers who rely on traffic from the world's market leading search engine, has been given counselling about its responses to queries.
Some of the results returned by the digital giant’s AI Gemini, designed to provide answers to search queries rather than links, have been far from clear.
Users reported bizarre and inaccurate answers to queries, some of them on the strange side such as: “How many rocks should I eat?”
The reported answer: Geologists recommend one rock per day.
And if you want to make cheese stick to the top of pizza, use non toxic glue.
News publishers are worried referral traffic from links to news appearing in search results will dry up, along with advertising revenue, as the new AI Summaries spread from the US to other countries.
However, Google insists that AI Overviews, as a jumping off point to visit web content, has been providing clicks of a higher quality to webpages.
“People are more likely to stay on that page, because we’ve done a better job of finding the right info and helpful webpages for them,” said Liz Reid VP, head of Google Search.
Google’s latest work in AI is seen by analysts as part of a global arms race to be the dominant player and protect its billions of dollars in advertising revenue.
The global digital platform, with the biggest share of the advertising market, agrees that recently odd, inaccurate or unhelpful AI Overviews have appeared in searches.
“And while these were generally for queries that people don’t commonly do, it highlighted some specific areas that we needed to improve,” said Liz Reid.
“One area we identified was our ability to interpret nonsensical queries and satirical content.”
In response, Google has made more than a dozen technical improvements.
These include better detection mechanisms for nonsensical queries that shouldn’t show an AI Overview, and limiting the inclusion of satire and humour content.
And limiting the use of user-generated content in responses that could offer misleading advice.
Google says it has also been vigilant in monitoring feedback and external reports, and taking action on the small number of AI Overviews that violate content policies.
“This means overviews that contain information that’s potentially harmful, obscene, or otherwise violative,” said Liz Reid.
“We found a content policy violation on less than one in every seven million unique queries on which AI Overviews appeared.”
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.