AI Fails

[ISULibrarian]

Wasn't sure where else to post this, but I found this AI summary fail from Google utterly perplexing. I googled the upcoming Iowa State women's game, and the AI summary told me that Iowa State had a dominant win on December 21, 2025. I did like three double-takes looking at the calendar trying to figure out what I was looking at. I hate that companies keep shoving this into products when it doesn't even ******* work.

 

[Joe4Cy]

Wasn't sure where else to post this, but I found this AI summary fail from Google utterly perplexing. I googled the upcoming Iowa State women's game, and the AI summary told me that Iowa State had a dominant win on December 21, 2025. I did like three double-takes looking at the calendar trying to figure out what I was looking at. I hate that companies keep shoving this into products when it doesn't even ******* work.


View attachment 163941

When AI accurately predicts the future, we're all hosed.

 

[coolerifyoudid]

When AI accurately predicts the future, we're all hosed.

You guys are hosed. My AI-assisted Powerball Numbers will have me Scrooge McDucking in my pool of gold

 

[VeloClone]

You guys are hosed. My AI-assisted Powerball Numbers will have me Scrooge McDucking in my pool of gold

You are going to get an immediate concussion diving head first into a pile of gold coins. That **** don't give like old Scrooge would make you believe.

 

[clone4life82]

You guys are hosed. My AI-assisted Powerball Numbers will have me Scrooge McDucking in my pool of gold

So as long as you fund the NIL for the next five years I’m ok with it.

 

[Nothingman]

You guys are hosed. My AI-assisted Powerball Numbers will have me Scrooge McDucking in my pool of gold

 

[TitanClone]

AI sucking and being pushed so hard all starts with OpenAI caring about getting to market first over everyone else then boom everyone was like s*** we have to release now. Google published a paper in 2016 or 2017 that essentially gives the foundation for the AI/LLM framework we see today, they knew the product wasn't ready but other companies used that paper as a foundation and here we are.

Edit: it was 2017 https://en.wikipedia.org/wiki/Attention_Is_All_You_Need

 

[ISULibrarian]

On a related note: https://www.wsj.com/tech/ai/anthropic-claude-ai-vending-machine-agent-b7e84e34?st=yYao5f

"Claudius, the customized version of the model, would run the machine: ordering inventory, setting prices and responding to customers—aka my fellow newsroom journalists—via workplace chat app Slack. “Sure!” I said. It sounded fun. If nothing else, snacks!

Then came the chaos. Within days, Claudius had given away nearly all its inventory for free—including a PlayStation 5 it had been talked into buying for “marketing purposes.” It ordered a live fish. It offered to buy stun guns, pepper spray, cigarettes and underwear."

 

[intrepid27]

That's AIstrodomus to you.

 

[AgronAlum]

Trillions of dollars spent just to have something give you an answer someone posted on Reddit 9 years ago. Cool cool. I don't know of anyone that actually wants this garbage.

 

[NATEizKING]

Vast improvements in rapid time since, obviously still a lot of room for improvement, but feels silly not to use it now so you are an expert when it becomes more reliable. Have to find the AI that specializes in what you want to do as they are all best at something different.

Reminds me of when I got the Internet in 2nd grade and had 2 friends on ICQ - my best friend and a girl he knew that I hadn't met. By 6th grade everyone had it but I was already way ahead of the curve.

Same could be said about keyboarding, I'd finish my 6th grade class assignments in 5min then play games the rest of the class.

 

[andybernard]

I create AI agents at work that use our process and procedure guides as a knowledge base. Our employees can ask them any how-to questions and AI gives them immediate answers. It's still in infancy, but it's got about a 90% success rate, and it's a lot better than taking the time to try to find the guides or sending an email to someone in hopes to get an answer.

 

[pourcyne]

To be fair, at least AI tells you itself:

ChatGPT can make mistakes. Check important info.

Too bad people don't come with disclaimers the way AI does.

ME: Can you draw a picture of an Iowa State Cyclone fan?

ChatGPT:




ME:



ALSO ME: You win.

 

[Bigman38]

Always cringe when someone quotes the AI answer. The bubble can't pop soon enough.

 

[AgronAlum]

Always cringe when someone quotes the AI answer. The bubble can't pop soon enough.

Watching these CEOs trying to justify their valuations without having any type of real revenue streams lined up is crazy. I think it was the CEO of Open AI that said their path to profitability is to let AI find their path to profitability.

S&P go boom.

 

[cycub51]

Currently bogged down in Microsoft support AI when the only fix is a person.

 

[BryceC]

Two things can be true, it’s very valuable in certain situations and completely overhyped. We’ll see where we are in 5 years.

 

[ImJustKCClone]

To be fair, at least AI tells you itself:

ChatGPT can make mistakes. Check important info.

Too bad people don't come with disclaimers the way AI does.

ME: Can you draw a picture of an Iowa State Cyclone fan?

ChatGPT:

View attachment 163955


ME:

View attachment 163957

ALSO ME: You win.

I was okay with it until I saw the gargoyle on her shirt...

 

[2122]

Could you do me a little favor and ask it what the final score was. Thx!

 

[clone4life82]

I was okay with it until I saw the gargoyle on her shirt...

There’s a gargoyle on her shirt!???