Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Oh stop. The neural network is piecing together the tokens in a way that indicates reasoning. Clearly. I don't really need to say this, we all know it now and so do you. Your statement here is just weak.

It's really embarrassing the stubborn stance people were taking that LLMs weren't intelligent and wouldn't make any progress towards agi. I sometimes wonder how people live with themselves when they realize their wrong outlook on things is just as canned and biased as the hallucinations of LLMs themselves.



Personal attacks really make your argument stronger.


I said your statement was weak I didn't say YOU were weak. Don't take it personally. It wasn't meant to be that way. If you take offense then I apologize.

That being said, my argument was a statement about a general fact that is very true. The sentiment not too long ago was these things are just regurgitators with no similarity to human intelligence.

I think it's clear now that all the previous claims were just completely baseless.


> That being said, my argument was a statement about a general fact that is very true.

Is this a general fact that is very true? It sounds like you are judging rather than stating a fact.

> It's really embarrassing the stubborn stance people were taking that LLMs weren't intelligent and wouldn't make any progress towards agi. I sometimes wonder how people live with themselves when they realize their wrong outlook on things is just as canned and biased as the hallucinations of LLMs themselves.


Yeah it is. Judgments can be true. Where is my judgement not true? It’s embarrassing to be so insistent on being right then finding out they’re so wrong.

Also I made a broad and general statement. I never referred to you or anyone personally here. If you got offended it only meant the judgement correctly referred to you and that you fit the criteria. But doesn’t that also make what I said true?


> If you got offended it only meant the judgement correctly referred to you

If you say "I hate all these black criminals I see everywhere" people will get offended even if they aren't black criminals, so your reasoning here is flawed.

And note how that is true even if there are a lot of black criminals around.

Why do people get upset over that? Its because they feel you judge more people as criminals than actually are. And its the same with your statement there, the way you phrased it it feels like you judge way more people than actually fit your description. If you want people to not judge you that way then you have to change how you write.


Right but HN is above that. Or so I thought? I thought this was a forum that people intelligently debate concepts and are able to maturely separate that from emotions. Are you saying you’re not above that? That if someone points something out that’s true about you that you don’t like to hear you get offended and emotional?

If that is the case then I apologize to you for saying something that is true about you and you don’t like to hear or face. Apologies, truly.


I don't think anyone has said LLMs wouldn't make any progress toward AGI, especially of researchers in the field. But a small piece of progress toward AGI is not the same as AGI.


Reasoning is a human, social and embodied activity. TFA is about machines that output text reminiscent of the results of reasoning, but it's obviously fake since the machine is neither human, social or embodied.

It's an attempt at fixing perceived problems with the query planner in an irreversibly compressed database.


That’s not what reason is. There’s nothing inherently human about reason.


I'm afraid it's not so clear. There are different perspectives on this.

For one, apart from humans, some animals, and now LLMs, there are but few entities that are able to apply reasoning. It may well be that reason is something that exists universally, but empirically this sounds a bit unlikely.


There's a sort of existential dread once man has created a machine that can perform on par with man himself. We don't want to face the truth so we move the goal posts and definitions.

To reason is now a human behavior. We moved the goal posts so we don't have to face the truth that AI crossed a barrier with the creation of LLMs. It doesn't really matter, there's no turning back now.


We can create beings "on par with man himself". If you ask around where you live someone will likely be able to show you some small ones and might perhaps introduce you to the activity that brings them to life.


Right, obviously man/child is on par with man himself.

I’m obviously talking about something on par with man himself but isn’t a man or even human. It is something else.


A non-human that can eat and gossip and make babies? What is this something?


A non human that is more intelligent than you and can outthink you and understand you better than you. That’s where the trendline is pointing.


I think the perspective is diametrically opposite to what you’re suggesting. It’s saying things that human do are not singular or sacrosanct. It’s a full acceptance that humanity will be surpassed.


Some people used to believe that. They imagined reason to be divine, hence the prime status of Aristotle and reason in roman catholicism.

But God is dead and has been for some time. You can try to change that if you want, but hitherto none of the prominent professional theologians have managed to wrestle with modern physics, feminist critique, the philosophies of suspicion or capitalism itself and reached a convincing win. Maybe you're the one, you should try and perhaps you'll at least become a free spirit.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: