Article The Feds are investigating ChatGPT's hallucination problem

[C C]

https://gizmodo.com/chatgpt-openai-ftc-i...1850636374

EXCERPTS: The Federal Trade Commision is investigating OpenAI to determine whether or not its ChatGPT AI chatbot and other products violate consumer protection laws by putting people’s personal reputations and data at risk.

[...] the FTC letter demands OpenAI provide a description of complaints it’s received to this point of its system making “false, misleading, disparaging or harmful” statements about people. Researchers and journalists have repeatedly shown examples of ChatGPT and other LLMs “hallucinating” fabricated information against subjects.

In an example first reported by Gizmodo, ChatGPT allegedly inserted a talk show host into an embezzlement court case he had nothing to do with. That radio host is now suing OpenAI for libel. The FTC says it wants to know the steps OpenAI took to filter or anonymize personal information included in its dataset and the steps it’s taken to decrease the chance its models conjure up fabricated statements about people.

The FTC also wants to know more about OpenAI’s policies and procedures for assessing safety and risk before releasing its products to the public... (MORE - missing details)

[Zinjanthropos]

If AI can’t be trusted then what? Be kind of like the internet today but without the billions of pages? Will science have to test all AI answers to establish truth?

AI tells Russia how to win war but in reality it’s really setting the Russians up for defeat…that would be interesting….one possible scenario?

[C C]

If AI can’t be trusted then what? Be kind of like the internet today but without the billions of pages? Will science have to test all AI answers to establish truth?

AI tells Russia how to win war but in reality it’s really setting the Russians up for defeat…that would be interesting….one possible scenario?

Surely LLMs can be infected with malware and turned into covert propaganda machines. And special interest groups and movements like Scientology will be placing their own versions before the public. Regardless, the world is headed back to the days before even the Enlightenment and Renaissance. The offshoots of postcolonialism and its grievances are purging Western standards.

What Was the Fact?
https://www.thenewatlantis.com/publicati...s-the-fact

EXCERPTS: This may sound odd. Facts are such a familiar part of our mental landscape today that it is difficult to grasp that to the premodern mind they were as alien as a filing cabinet. But the fact is a recent invention.

[...] The rise of facts was the triumph of a certain kind of shared empirical evidence over personal experience, deduction from preconceived ideas, and authoritative diktat. Facts stand on their own, free from the vicissitudes of anyone’s feelings, reasoning, or power.

[...] Centuries ago, our society buried profound differences of conscience, ideas, and faith, and in their place erected facts, which did not seem to rise or fall on pesky political and philosophical questions. But the power of facts is now waning, not because we don’t have enough of them but because we have so many. What is replacing the old hegemony of facts is not a better and more authoritative form of knowledge but a digital deluge that leaves us once again drifting apart.

As the old divisions come back into force, our institutions are haplessly trying to neutralize them. This project is hopeless — and so we must find another way. Learning to live together in truth even when the fact has lost its power is perhaps the most serious moral challenge of the twenty-first century... (MORE - missing details)

[stryder]

If AI can’t be trusted then what? Be kind of like the internet today but without the billions of pages? Will science have to test all AI answers to establish truth?

AI tells Russia how to win war but in reality it’s really setting the Russians up for defeat…that would be interesting….one possible scenario?

Surely LLMs can be infected with malware and turned into covert propaganda machines. And special interest groups and movements like Scientology will be placing their own versions before the public. Regardless, the world is headed back to the days before even the Enlightenment and Renaissance. The offshoots of postcolonialism and its grievances are purging Western standards.

What Was the Fact?
https://www.thenewatlantis.com/publicati...s-the-fact

EXCERPTS: This may sound odd. Facts are such a familiar part of our mental landscape today that it is difficult to grasp that to the premodern mind they were as alien as a filing cabinet. But the fact is a recent invention.

[...] The rise of facts was the triumph of a certain kind of shared empirical evidence over personal experience, deduction from preconceived ideas, and authoritative diktat. Facts stand on their own, free from the vicissitudes of anyone’s feelings, reasoning, or power.

[...] Centuries ago, our society buried profound differences of conscience, ideas, and faith, and in their place erected facts, which did not seem to rise or fall on pesky political and philosophical questions. But the power of facts is now waning, not because we don’t have enough of them but because we have so many. What is replacing the old hegemony of facts is not a better and more authoritative form of knowledge but a digital deluge that leaves us once again drifting apart.

As the old divisions come back into force, our institutions are haplessly trying to neutralize them. This project is hopeless — and so we must find another way. Learning to live together in truth even when the fact has lost its power is perhaps the most serious moral challenge of the twenty-first century... (MORE - missing details)

I'm pretty sure the QAnon movement was a Chatbot. It just hallucinated certain "facts" and people seemed to lap it up.

The problem is that if you get a Chatbot acting as an influencer, it can be operated by bad actors that use those that drink the Q-aid to their own advantage.

[confused2]

We have to accept that all or any of what follows may be false...

The parties involved in the hallucination story all seem to be very 'pro gun' - for example SAF is Second Amendment Foundation, Walters seems to have received a Gun Rights Defender of the Year award from SAF and Riehl seems to be Editor-In-Chief & Owner of AmmoLand.com.

All nice people so why are they up in arms when one [of them] or more are the the victim of unintentional damage [to their reputation]? Accidents will always happen - that doesn't make guns (or AI) a bad thing.

[Zinjanthropos]

I would think that AI isn’t going to be hallucinating. IOW there’s method to its madness or somebody with a brain is behind it. Maybe it was a glitch. Regardless, not going to be much longer until AI goes to work for the bad guys if it isn’t already. Blackberry is a Canadian Cybersecurity company and here’s what they anticipate…

https://blogs.blackberry.com/en/2023/02/...e-imminent

Q: Will AI know right from wrong? Most criminals run from scene of the crime, so people know for the most part. Will AI consciously break laws? If AI hacks into another computer will it be able to chat with the victimized machinery and learn, even possibly reject an illegal hack?