Google's AI Overview is a dangerous disaster
The new AI feature from Google is spewing out misinformation
đ¤ Googleâs AI is spreading misinformation
đŽ The companyâs new AI Overview is providing some bizarre answers
đ¤Śââď¸ Google recommends putting glue on your pizza
đś It also claimed that a dog played in the NBA
Google used to be the home of facts, truth, and information, settling any debate within a few minutes. âGoogle it,â people used to say.
However, the search giantâs pursuit of AI has led to one of the most valuable resources of information becoming about as useful as a chocolate fireguard, and genuinely dangerous too.
The companyâs new AI Overview summaries, which accompany traditional searches, are spewing out misinformation. When asked âcheese not sticking to pizzaâ, Googleâs AI Overview recommends using glue. Seriously.
âAdd some glue,â says Google. âMix about 1/8 cup of Elmerâs glue in with the sauce. Non-toxic glue will work.â
Googleâs AI answer, which we strongly recommend no one follows for obvious reasons, has been pulled from a comment from a user named âfucksmithâ in a decade-old Reddit thread, who was clearly joking about using glue on a pizza.
However, this isnât the only mistake users have seen from Googleâs AI Overview. Others have included that a cat licks your hands to see whether youâre fit for consumption, a dog playing in the NBA, and other mishaps that only serve to give people the wrong information.
Google spokesperson Meghann Farnsworth said the mistakes appear from âgenerally very uncommon queries, and arenât representative of most peopleâs experiences.â (thanks, The Verge). Google also includes a label that its âGenerative AI is experimentalâ at the bottom of any given answer.
But in a world where the truth has been replaced with âmy truthâ and fake news is rife, itâs clear Googleâs AI Overview isnât ready for the masses. It also begs the question as to where Google is obtaining its information.
Reddit has become a particular favorite with Googleâs search ranking algorithm, with discussions usually taking pride of place at the top of a page. But as Reddit is filled with just regular everyday folk, and not usually experts in their field, it sets a dangerous precedent.
AI may be the latest and greatest thing in the tech industry right now, but if itâs currently serving to make a product that millions of people rely on and trust worse, is it really the future?
This wonât be the last time we see AI provide a bizarre answer, but Google needs to take extra precautions to stop the spread of misinformation before people look elsewhere to get the answers theyâre after.
Adam Vjestica is The Shortcutâs Senior Editor. Formerly TechRadarâs Gaming Hardware Editor, Adam has also worked at Nintendo of Europe as a Content Marketing Editor where he helped launch the Nintendo Switch. Follow him on X @ItsMrProducts.