Live data from Hacker News

Creativity has left the chat: The price of debiasing language models

arxiv.org

211–220 of 238 posts

Re: Creativity has left the chat: The price of debiasing language models

#211
post #207
post #83

Earlier quoted context omitted.

I guess our disagreement here is that I don't think AIs are moral entities/are capable of being harmed or that training AIs and teaching humans are comparable. Being abusive to pupils isn't wrong because of something fundamental across natural and machine learning, it's wrong because it's harmful to the pupils. In what way is it possible to harm an LLM?

> I don't think AIs are moral entities/are capable of being harmed or that training AIs and teaching humans are comparable. Notice how this is a completely different argument that has nothing in common with what you originally said - "I don't understand the take that training an AI is an amoral act but aligning an AI is inherently moral. They're exactly the same, processes for adjusting parameters to get a desired ou…

That's pretty uncharitable. You pivoted the conversation by introducing a new hypothetical for me to respond to. Of course my response is different. There's no conflict between the two comments.

If we're going to be play that game, notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable? I actually engaged substantively with your hypothetical; are you able to do the same?

Re: Creativity has left the chat: The price of debiasing language models

#212
The official openai-cookbook (https://github.com/openai/openai-cookbook) used to have an explicit, but buried, call out that instruction-following models like `text-davinci-003` were "Less diverse; less creative; sometimes harder to steer tone, style, etc." as opposed to base completion models like `davinci`.

It stood out to me because it seemed to be an internal admission that this training narrowed the potential of the models.

Required a bit of digging but I found the old file in the history, the relevant text is in the comparison table at the bottom: https://github.com/openai/openai-cookbook/blob/c651bfdda64ac...

Re: Creativity has left the chat: The price of debiasing language models

#213
Distilling my thoughts on 'debiasing' here, and in a variety of other modern endeavors.

It is better to have representations of reality that you can then discuss and grapple with honestly, than to try to distort representations - such as AI - to make them fit some desired reality and then pressure others to conform their perception to your projected fantasy.

Representations don't create reality, and trying to use representations in that way only causes people to go literally insane, and to divide along lines of who accepts and who rejects your fantasy representation.

So, for example, if you try and remove any racial bias from AI, you are going to end up crushing the AI's ability to represent reality according to a variety of other real factors: income, judicial outcomes, health risks, etc. Your desired reality makes the actual tool worthless, except to confirm one group's own intended fantasy world as they envision it. The problem doesn't get dealt with, it just becomes impossible to think about or discuss.

So instead of dealing with real problems, you hope you can simply prevent people from thinking thoughts that cause those problems by wrapping them in a bubble that deflects those thoughts before they happen. This is magical, wizardry thinking: treating words as if they create reality, instead of merely describing it. And it will break, eventually, and in a very ugly way: people dividing along lines of their perception of reality, even more than they already do.

Re: Creativity has left the chat: The price of debiasing language models

#214
post #207

Earlier quoted context omitted.

> I don't think AIs are moral entities/are capable of being harmed or that training AIs and teaching humans are comparable. Notice how this is a completely different argument that has nothing in common with what you originally said - "I don't understand the take that training an AI is an amoral act but aligning an AI is inherently moral. They're exactly the same, processes for adjusting parameters to get a desired ou…

That's pretty uncharitable. You pivoted the conversation by introducing a new hypothetical for me to respond to. Of course my response is different. There's no conflict between the two comments. If we're going to be play that game, notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable? I actually engaged substantively with your…

> You pivoted the conversation by introducing a new hypothetical for me to respond to.

I wasn't trying to introduce anything new, I was trying to point out a gap in the logic of your original statement.

> notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable?

Yes, of course, I wrote that to explain why I'm not engaging on this new, different claim.

Re: Creativity has left the chat: The price of debiasing language models

#215
post #214

Earlier quoted context omitted.

That's pretty uncharitable. You pivoted the conversation by introducing a new hypothetical for me to respond to. Of course my response is different. There's no conflict between the two comments. If we're going to be play that game, notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable? I actually engaged substantively with your…

> You pivoted the conversation by introducing a new hypothetical for me to respond to. I wasn't trying to introduce anything new, I was trying to point out a gap in the logic of your original statement. > notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable? Yes, of course, I wrote that to explain why I'm not engaging on this n…

The nerve of me, to expand on my views as a discussion develops. Of course you have lots of great points to make, but you can't share them with the likes of me.

Re: Creativity has left the chat: The price of debiasing language models

#216
post #214

Earlier quoted context omitted.

> You pivoted the conversation by introducing a new hypothetical for me to respond to. I wasn't trying to introduce anything new, I was trying to point out a gap in the logic of your original statement. > notice how you didn't actually respond to my comment or explain why you thought LLMs were moral entitles or why ML and teaching were comparable? Yes, of course, I wrote that to explain why I'm not engaging on this n…

The nerve of me, to expand on my views as a discussion develops. Of course you have lots of great points to make, but you can't share them with the likes of me.

> The nerve of me, to expand on my views as a discussion develops.

Nothing wrong with expanding your views. But you've neither defended nor retracted your original argument. I'm trying to stick to that.

> Of course you have lots of great points to make, but you can't share them with the likes of me.

I don't have anything to say about your new argument (which may be great and compelling), I haven't thought through it at all, I'm trying to avoid getting sidetracked.

Re: Creativity has left the chat: The price of debiasing language models

#217
post #71
post #32

Something I notice about text written by LLMs is how painfully obvious they are to identify sometimes. Recently I was watching a very well researched two hour video on Tetris World Records [1], but the sheer amount of text clearly "enhanced" by an LLM really made me uncomfortable. ChatGPT speaks a very specific, novel, dialect of English, which I've come to deeply despise. I'd always guessed it was caused by some kin…

> ChatGPT speaks a very specific, novel, dialect of English, which I've come to deeply despise. There was this article saying that ChatGPT output is very close to the Nigerian business english dialect, because they hired a lot of people from there. Might have even been posted on HN.

https://www.theguardian.com/technology/2024/apr/16/techscape...

Re: Creativity has left the chat: The price of debiasing language models

#218
post #216

Earlier quoted context omitted.

The nerve of me, to expand on my views as a discussion develops. Of course you have lots of great points to make, but you can't share them with the likes of me.

> The nerve of me, to expand on my views as a discussion develops. Nothing wrong with expanding your views. But you've neither defended nor retracted your original argument. I'm trying to stick to that. > Of course you have lots of great points to make, but you can't share them with the likes of me. I don't have anything to say about your new argument (which may be great and compelling), I haven't thought through it…

Let's just be perfectly clear here.

You asked: why should we treat these two cases differently?

I answered: one of them involves harm and one can't possibly.

You then refused to discuss the topic further. I never changed my mind. I never contradicted myself. I have nothing to retract. I simply elaborated on my views in response to your question. Regardless of whether you "didn't intend to introduce something new" - you did, you introduced a new hypothetical and a new challenge.

I don't know why you are reacting this way, but let's be perfectly clear, I engaged substantively with your points, and you simply refused to continue the discussion. That's all well and good, you're under no obligation. But you're saying I've done something wrong here, and I really haven't.

Frankly I find the whole thing bizarre.

Re: Creativity has left the chat: The price of debiasing language models

#219

Distilling my thoughts on 'debiasing' here, and in a variety of other modern endeavors. It is better to have representations of reality that you can then discuss and grapple with honestly, than to try to distort representations - such as AI - to make them fit some desired reality and then pressure others to conform their perception to your projected fantasy. Representations don't create reality, and trying to use rep…

"Reality" is a tricky concept. For me, I follow Jeff Atwood - if it isn't written down, it doesn't exist. According to this logic, people wasted a lot of time on imaginary, illusory things for most of human history, but now they have phones and most communication is digital so there is the possibility to finally be productive. This definition shows how the concept of distorting reality or honestly representing reality is flawed - reality is what I write down, I can in fact create more reality by writing down words, and regardless of what I write, it will be reality. Representations like books, scrolls, papyri constitute the reality of most civilizations - there is no other evidence they existed. It is true that representations don't create reality - rather, humans create representations, and these representations collectively are reality, no creation involved.

Representations are art - for example books, they are "literary art". It is uncontroversial that people will like and dislike certain works. It is more controversial whether art can be "inherently" good or bad. PG actually wrote an essay, https://www.paulgraham.com/goodart.html, arguing that there is a meaningful metric, and that one can learn how to have good taste, defined as being able to identify whether the work is universally appealing or distasteful to humanity. There is good art and people will notice if it is good. I think this is uncontroversial in the LLM space, there are various benchmarks and human rating systems and people have formed a rough ranking of models. Now when there is good art, there is also bad. And similarly bad representations. There is a myth that representations can make people insane - for example, the concept of infinity, or NSFL images - but practically, words can't hurt you. You can make and break representations with abandon and nothing will happen, other than wasting your time. It is just that some representations are bad. Like phlogiston, aether, ... complete dead ends. Trust me when I say you will read the Wikipedia page and come away wondering why the ancients were so stupid. That is all trying to remove racial bias is, is improving art. Whether it crushes the AI's ability or not is a matter of science and taste, and so far experiments have been promising.

To focus on exactly why your perspective is misguided: Can you describe what there is about reality that cannot be described with words? :-)

Re: Creativity has left the chat: The price of debiasing language models

#220
post #98

Earlier quoted context omitted.

You wrote further above that "I don't understand the notion", and that was spot on. Should've stopped there rather than here, in my opinion, but feel free to disagree.

I'm sorry if anything I said insulted you or seemed to be a comment on you personally. That wasn't my intention.

Apology accepted and please accept mine in return, that was much too spiky a reply from me.
Post reply on HN