Live data from Hacker News

We need to tell people ChatGPT will lie to them, not debate linguistics

simonwillison.net

421–430 of 485 posts

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#421

The part that's concerning about ChatGPT is that a computer program that is "confidently wrong" is basically indistinguishable from what dumb people think smart people are like. This means people are going to believe ChatGPT's lies unless they are repeatedly told not to trust it just like they believe the lies of individuals whose intelligence is roughly equivalent to ChatGPT's. Based on my understanding of the appro…

Many of the smarter people are still wrong on what happened on many topics in 2020. They were fooled by various arguments that flew in the face of reality and logic because fear and authority was used instead.

The people that avoid this programming isn't based on smart or stupid. It's based on how disagreeable and conscientious you are. A more agreeable and conscientious person can be swayed more easily by confidence and emotional appeals.

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#422

Earlier quoted context omitted.

What's terrible about some Wikipedia pages is when you personally don't know they are controversial. It should be vastly more prominent when there are tensions on an article and even admins shouldn't have the power to hide this.

Admins do not, in fact, have the power to hide tensions on an article, as even the act of deletion produces public logs. Certainly if you don't know to look for them, you won't see them, though.

The main page should become a disjointed piece of both sides writing areas of the article. Right now, admins pick the side and moderate the other away into the history.

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#423

Earlier quoted context omitted.

To clear this up, because open Holocaust denial is never OK: Thousands of jews were indeed murdered at Babi Yar, you can find their names online [0]. Paul Blobel [1] was executed at Nuremberg for this. No serious source disputes the killing of jews at Babi Yar, be it the US, Germany (where this matter went to court), Israel, Ukraine, or Russia. 0: https://yvng.yadvashem.org/index.html?language=en&s_id=&s_la... 1: htt…

I couldn't find any connection between the names on the list and the people who lived in Kiev during the war. Prior to the war, there were 224,000 Jews living in Kiev, and approximately 200,000 were evacuated to the east. Around 40,000 Jews remained, including my grandmother. This is a relatively small number of people that can be verified.

"You couldn't find any connection" but who cares what you claim you did?

You asserted Babi Yar was a "Russian fake" but no sane country in the world fails to acknowledge it happened and it was a huge massacre of Jews.

What some weirdo Holocaust denier claims on the internet doesn't matter.

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#424

Earlier quoted context omitted.

"simply having more data doesn’t turn a machine that can’t reason into one that can" But that's exactly what's been happening. LLM's are a triumph of "big data" used by dumb algorithms to create a reasoning machine.

A reasoning machine that can't reason, makes up fake evidence and doesn't follow simple logic.

Are we talking about humans or..

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#425
post #398

Earlier quoted context omitted.

[flagged]

What on Earth are you about? Babi Yar is one of the single largest massacres of Jews during WW2. Some 30,000 Jews died in the first two days of the massacre. The overall SS commander in charge, Paul Blobel, was sentenced to death at Nuremberg. The event is widely remembered as a massacre of Jews. It's acknowledged by Ukraine as well. This is not "a Russian fake" and there is plenty of evidence for it. Please, don't r…

As I can see in Nuremberg Trials, Babi Yar was mentioned twice:

18 October 1945, Berlin

In the Crimea peaceful citizens were gathered on barges, taken out to sea and drowned, over 144,000 persons being exterminated in this manner.

In the Soviet Ukraine there were monstrous criminal acts of the Nazi conspirators. In Babi Yar, near Kiev, they shot over 100,000 men, women, children, and old people. In this city in January 1942, after the explosion in German Headquarters on Dzerzhinsky Street the Germans arrested as hostages 1,250 persons—old men, minors, women with nursing infants. In Kiev they killed over 195,000 persons.

In Rovno and the Rovno region they killed and tortured over 100,000 peaceful citizens.

20 November 1945, Nuremberg:

In the Crimea peaceful citizens were gathered on barges, taken out to sea and drowned, over 144,000 persons being exterminated in this manner.

In the Soviet Ukraine there were monstrous criminal acts of the Nazi conspirators. In Babi Yar, near Kiev, they shot over 100,000 men, women, children, and old people. In this city in January 1941, after the explosion in German headquarters on Dzerzhinsky Street the Germans arrested as hostages 1,250 persons—old men, minors, women with nursing infants. In Kiev they killed over 195,000 persons.

In Rovno and the Rovno region they killed and tortured over 100,000 peaceful citizens.

Where can I find more information about the Babyn Yar crime that was presented on the Nuremberg Trials?

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#426
post #394

Earlier quoted context omitted.

> A lot smart people seem to be saying that the existing approaches have room to improve simply by training with more data. There’re also many “smart people” saying that simply having more data doesn’t turn a machine that can’t reason into one that can.

"simply having more data doesn’t turn a machine that can’t reason into one that can" But that's exactly what's been happening. LLM's are a triumph of "big data" used by dumb algorithms to create a reasoning machine.

> dumb algorithms

I’m not an expert and haven’t seen architecture of ChatGPT. But it’s probably far from dumb. You can’t simply put more data into any algorithm and expect improvement. You need a model to be able to take advantage of the data and train in reasonable time. That’s exactly what’s been happening in ML field.

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#427
post #423

Earlier quoted context omitted.

I couldn't find any connection between the names on the list and the people who lived in Kiev during the war. Prior to the war, there were 224,000 Jews living in Kiev, and approximately 200,000 were evacuated to the east. Around 40,000 Jews remained, including my grandmother. This is a relatively small number of people that can be verified.

"You couldn't find any connection" but who cares what you claim you did? You asserted Babi Yar was a "Russian fake" but no sane country in the world fails to acknowledge it happened and it was a huge massacre of Jews. What some weirdo Holocaust denier claims on the internet doesn't matter.

[flagged]

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#428
post #224
post #154

Earlier quoted context omitted.

> but why would you expect that from it? If you understand what it is doing, then you don't. But the layman will just see a computer that talks in language they understand, and will infer intent and sentience are behind that, because that's the only analog they have for a thing that can talk back to them with words that appear to make sense at the complexity level that ChatGPT is achieving. Most humans do not have su…

> Most humans do not have sufficient background to understand what they're really being presented with, they will take it at face value. Trust in all forms of media [1], and institutions [2], is at an all time low. I'm not sure why that distrust would go away, with reading the output from a company that's clearly censuring/fudging that output. I think any "damaging" trust would, clearly, be a transitory phenomenon, s…

[flagged]

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#429
post #117
post #52

I think characterisation of LLMs as lying is reasonable because although the intent isn't there to misrepresent the truth in answering the specific query, the intent is absolutely there in how the network is trained. The training algorithm is designed to create the most plausible text possible - decoupled from the truthfulness of the output. In a lot of cases (indeed most cases) the easiest way to make the text plaus…

> The training algorithm is designed to create the most plausible text possible - decoupled from the truthfulness of the output. In a lot of cases (indeed most cases) the easiest way to make the text plausible is to tell truth. Yes. > But guess what, that is pretty much how human liars work too! There is some distinction between lying and bullshit. https://en.wikipedia.org/wiki/On_Bullshit#Lying_and_bullshit

Yes, this is exactly it. ChatGPT is superhuman at the task of generating bullshit.

Re: We need to tell people ChatGPT will lie to them, not debate linguistics

#430
post #423

Earlier quoted context omitted.

"You couldn't find any connection" but who cares what you claim you did? You asserted Babi Yar was a "Russian fake" but no sane country in the world fails to acknowledge it happened and it was a huge massacre of Jews. What some weirdo Holocaust denier claims on the internet doesn't matter.

[flagged]

It's your fault that you deny the widely accepted Babi Yar massacre and call it "a Russian fake", exactly what Holocaust deniers do. If the shoe fits...

So in your opinion, countries such as all of Europe (Germany included), the US, Israel, etc are all part of this conspiracy?

Post reply on HN