Does it worry you that GPT might give inaccurate or misleading information?

[Bootstrap]

This poll is about the accuracy of GPT. For the purposes of this poll, GPT refers to any system like OpenAI, Grok, ChatGPT, Claude, NLLB, etc.

[Ken]

I'm not so concerned about benign use like looking up basic facts on the internet. That sort of thing is really low stakes. Although obviously GPT frequently produces complete garbage answers.

I'm far more worried about the use of AI for decisions that have real life consequences. Such as companies using AI to deny insurance. Or the government using AI for sentencing. Or to determine guilt or innocence. Or to determine who lives or dies in a war. Or for frankly any use that has real world consequences.

[Neto]

It annoys me that Google Search now places AI responses at the top of any search, and their policies now reduce the chance that I will receive reliable answers to questions from quotable sources.

[Just this morning I have been attempting to get an answer about placement of a flooring transition at the entry to a hallway. The automated responses automatically flip to placement of the transition where there is a door. (Under the door - that's logical.) Where for a hallway entrance? Centered? At the edge of the hallway, or at the edge of the living room? (Not wanting to divert this discussion here, so no response expected. Just explaining my frustration.)

The whole search process has become a vexation. In the past you might finally get an answer by looking through multiple 'pages' of 'hits', but now sometimes there is only a single page, or none at all, especially when you keep putting the beginning and ending quotation marks back in, to hopefully eliminate the completely irrelevant 'answers'.]

[Robert]

It annoys me that Google Search now places AI responses at the top of any search, and their policies now reduce the chance that I will receive reliable answers to questions from quotable sources.

[Just this morning I have been attempting to get an answer about placement of a flooring transition at the entry to a hallway. The automated responses automatically flip to placement of the transition where there is a door. (Under the door - that's logical.) Where for a hallway entrance? Centered? At the edge of the hallway, or at the edge of the living room? (Not wanting to divert this discussion here, so no response expected. Just explaining my frustration.)

The whole search process has become a vexation. In the past you might finally get an answer by looking through multiple 'pages' of 'hits', but now sometimes there is only a single page, or none at all, especially when you keep putting the beginning and ending quotation marks back in, to hopefully eliminate the completely irrelevant 'answers'.]

I have found the challenge is learning how to ask the question. AI does not think like us, so explaining so it understands the question does take some effort. I also not not accept its response totally. It is like another person, they could be right or wrong, or somewhere in between.

[ken_sylvania]

No, it doesn't worry me. I know it happens and is going to happen so what's the point of worrying about it.

[mike]

No. It is similar to worrying that the public library, or a Google search, or MennoNet, or the guys at the local diner will give you inaccurate or misleading information.

[barnhart]

Reasonable people will use it reasonably and there will always be those who use it otherwise. The areas that cause a little concern
are the ways it is used outside my control to make determinations that might effect me. For example deciding which medical conditions deserve insurance coverage for treatment or who goes in the high risk pool for auto insurance or which neighborhoods need or do not need law enforcement crackdowns. Or maybe for setting global tax policy. I saw someone who was able to reproduce the Trump tariff policy with AI by fine tuning a few prompts. It's possible an intern did something similar to produce a rough draft which became policy, at least for a day or two.

[JohnH]

Reasonable people will use it reasonably and there will always be those who use it otherwise. The areas that cause a little concern
are the ways it is used outside my control to make determinations that might effect me. For example deciding which medical conditions deserve insurance coverage for treatment or who goes in the high risk pool for auto insurance or which neighborhoods need or do not need law enforcement crackdowns. Or maybe for setting global tax policy. I saw someone who was able to reproduce the Trump tariff policy with AI by fine tuning a few prompts. It's possible an intern did something similar to produce a rough draft which became policy, at least for a day or two.

The AI that produced that (I saw the same thing) probably was using the Trump tariff policies reported in the media as training data.

[Neto]

I realize that this sort of thing has "come a long way" by now, but I'm reminded of what I was told by a dairy farmer (probably more than 40 years ago). A program had been developed to determine the amount of feed to give each cow, based on milk production. (Each cow wore a collar that tripped the system to give it the amount of feed determined by the program.) The dairy farmer told me that this program could starve a cow to death, because any slight reduction in yield on any given day would be automatically entered into the program's calculations, so the next feed time would be reduced, on and on, until the farmer noticed what was happening and strangled the program before it could kill his cow.

[Jazman]

I've seen this small related news... from the GOP big bill they are trying to get through Congress...

(1) IN GENERAL - Except as provided in paragraph (2), no State or political subdivision thereof may enforce any law or regulation regulating artificial intelligence models, artificial intelligence systems or automated decision systems during the 10-year period beginning on the date of the enactment of this Act.

Has anyone read/studied the GOP's (and or Trump's) big beautiful budget bill and can confirm this is really part of it?
If it is part of it, why would the GOP want to discourage any law or regulation having to do with Ai systems? Is it for their benefit somehow or do they think it will benefit the country at large and why? (Another source I scanned suggested the tech industry would prefer a national Ai regulations/standards vs a state-by-state approach... )