Get ready to start using Bing or DuckDuckGo for real this time. Googling things is now a fool's errand after the company launched its new AI Overview product in search.
Now, overtaking both regular search results and the ad placements, Google has began serving up summarizations of search queries, after striking deals worth millions of dollars with News Corp and Reddit for their data to be used as source material.
Many people have noticed that not only is there very little actual intelligence going on, the results simply regurgitate words from sources instead of relaying helpful information or analyzing the material. Within minutes, satirical Reddit comments were cited and sourced as legitimate answers, and don't be surprised if fake quotes from sites like The Onion appear in your search as gospel.
While the use of AI technology has been criticized — specifically large language models, which seem to have a very limited scope according to some researchers and just not enough data to learn from — the way its been rolled out by corporations has been disastrous.
It's now, unfortunately, up to users to find glaring issues and outright lies that pass through the cracks when it comes to AI-generated assisted scenarios. Google already messed up earlier this year with Gemini, their AI assistant, and now their larger, wide-scale AI overview assistance during search has met the same fate.
Here are some of the worst and most egregious examples of misinformation worst Twitter have found since the AI overview function was rolled out.
it's so cool that they fed every reddit shitpost into this thing and there's probably no way to fix it now pic.twitter.com/W0I0wjbeAx
— lauren (@Very__Regular) May 24, 2024
This reminds me of a much longer personal story of where a company I used to work at switched from human fact checkers to “crowd sourced algorithm” to deliver news. It never worked, made everything worse and ended up being a fancy buzzword to dupe C-level execs, just like AI! https://t.co/FjTFxv4B3B
— Scott Bromley (@Scott_Bromley) May 23, 2024
I'm learning a lot about American history with Google's AI Overview pic.twitter.com/37vmpepdHK
— Bobby Allyn (@BobbyAllyn) May 23, 2024
I tested this out and yes, when you Google "is poison good for you," the A.I. response not only says it's "beneficial to humans in a number of ways," it takes FIVE BULLET POINTS before it says ANYTHING cautionary.
— William Bibbiani (@WilliamBibbiani) May 23, 2024
And even then it only says "toxins can be bad in LARGE AMOUNTS." https://t.co/A0xszSzbIk pic.twitter.com/qeu09LYEZe
Tried this and got a slightly different overview but the worst part is when I scrolled down there was a story headlined “Geologists Recommend Eating At Least One Small Rock Per Day” from…The Onion https://t.co/pdSbuHMTDc
— Lauren Goode (@LaurenGoode) May 24, 2024
Meanwhile, over in Google Search.
— MMitchell (@mmitchell_ai) May 22, 2024
Andrew Johnson has been killin it, I never knew. pic.twitter.com/IV2zCmI6Zv
https://t.co/4kvJVLDRTb pic.twitter.com/827R4txYD2
— NamesAidan (@MyNamesAidan) May 22, 2024
Thank you Google AI pic.twitter.com/YAJvUuGyun
— 🏳️⚧️Graph Crimes🏳️🌈 (@GraphCrimes) May 24, 2024
"Yeah these queries are super uncommon which is why so many people are commonly finding them and posting them" https://t.co/hrV8TDPrW5
— Ed Zitron (@edzitron) May 23, 2024
I saw someone else mention this one on another post
— Raven - 🍉 (@RavenTenebris13) May 20, 2024
Ah yes , the snake very famous for being a mammal 🤣 pic.twitter.com/xo538FcqoK
Found one thing Google's new AI Overview definitely gets right pic.twitter.com/ixJZjCsquM
— Brian Merchant (@bcmerchant) May 23, 2024
Before Google, if you didn’t know something you had to go ask someone and most of the time they couldn’t help you, and now that’s also how Google works.
— Noah Garfinkel (@NoahGarfinkel) May 24, 2024
I feel the same dissonance seeing creatives say they're excited about AI as I had when (often the same) creatives raved about NFTs. At best they don't understand the limitations baked into the tech, at worst it's just chum for investors and excitement over team reduction.
— Mike Bithell (@mikeBithell) May 24, 2024
Feels like Google assumed that putting “This is experimental” in very small text on their AI results would make up for them being wrong all the time but that’s the kind of conclusion you come to when you only talk to other AI pod people for a solid year.
— chris person (@Papapishu) May 23, 2024
Kind of incredible how easy it is to identify what he means by each of these
— Kyle Orland (@kyleor.land) May 24, 2024 at 12:26 PM
[image or embed]
they call it AI because Ayyyy I dunno about this crap
— aLec robBins (@alecrobbins) May 23, 2024
