Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Sam Altman goes before US Congress to propose licenses for building AI

reuters.com

Post reply on HN
Print view View original

381–390 of 1001 posts

Re: Sam Altman goes before US Congress to propose licenses for building AI

#381

Post by Culonavirus »

srslack wrote: post #9

Imagine thinking that regression based function approximators are capable of anything other than fitting the data you give it. Then imagine willfully hyping up and scaring people who don't understand, and because it can predict words you take advantage of the human tendency to anthropomorphize, so it follows that it is something capable of generalized and adaptable intelligence. Shame on all of the people involved in…

> Imagine thinking that regression based function approximators are capable of anything other than fitting the data you give it.

Literally half (or more) of this site's user base does that. And they should know better, but they don't. Then how can a typical journo or a legislator possibly know better? They can't.

We should clean up in front of our doorstep first.

Re: Sam Altman goes before US Congress to propose licenses for building AI

#383

Post by Vox_Leone »

candiddevmike wrote: post #174

I'm sad that we've lost the battle with calling these things AI. LLMs aren't AI, and I don't think they're even a path towards AI.

>>I'm sad that we've lost the battle with calling these things AI. LLMs aren't AI, and I don't think they're even a path towards AI.

Ditto the sentiments. What about other machine learning modalities, like image detection? Will I need a license for my mask rcnn models?. Maybe it is just me, but the whole thing reeks of control

Re: Sam Altman goes before US Congress to propose licenses for building AI

#384

Post by sangnoir »

hellojesus wrote: post #375
sangnoir wrote: post #252

Earlier quoted context omitted.

I guess @sama took that leaked Google memo to heart ("We have no moat... and neither does OpenAI"). Requiring a license would take out the biggest competitive threats identified in the same memo (Open Source projects) which can result in self-hosted models, which I suppose Altman sees as an existential threat to OpenAI

There is no way to stop self hosted models. The best would be to send gov to data centers, but what if those centers are outside US jurisdiction? Too funny to watch the gov play these losing games.

> There is no way to stop self hosted models.

edit: Current models- sure, but they will soon be outdated. I think the idea is to strangle the development of comparable, SoTA models in the future that individuals can self-host; OpenAI certainly won't release their weights, and they'd want the act of releasing weights without a license to be criminalized. If such a law is signed, it would remove the threat of smaller AI companies from disintermediating OpenAI, and individuals from collaborating to engage in any activity that results in publicly available model weights (or even making the recipe itself illegal to distribute)

Re: Sam Altman goes before US Congress to propose licenses for building AI

#385

Post by bick_nyers »

JumpCrisscross wrote: post #265
brenns10 wrote: post #234

Reminds me of SBF calling for crypto regulations while running FTX. Being seen as friendly to regulations is great for optics compared to being belligerently anti-regulation. You can appear responsible and benevolent, and get more opportunity to weaken regulation by controlling more of the narrative. And hey, if you get end up getting some regulatory capture making competition harder, that's a great benefit too. Open…

> Reminds me of SBF calling for crypto regulations while running FTX Scott Galloway called it the stop-me-before-I-kill-grandma defence. (Paraphrasing.) You made money making a thing. You continue to make the thing. You’re telling us how the thing will bring doom and gloom if not dealt with (conveniently implying it will change the world). And you want to staff the regulatory body you call for with the very butchers…

Sure, I get it, but if Sam Altman quit tomorrow, would it stop Economic Competition -> Microsoft Shareholders -> Microsoft -> OpenAI?

Is there really a better alternative here?

Re: Sam Altman goes before US Congress to propose licenses for building AI

#386

Post by joaogui1 »

zootreeves wrote: post #189
rdedev wrote: post #146

Earlier quoted context omitted.

To be fair LLMs are predicting the next token. It's just that to get better and better predictions it needs to understand some level of reasoning and math. However it feels to me that a lot of this reasoning is brute forced from the training data. Like chatgpt gets some things wrong when adding two very large numbers. If it really knew the algorithm for adding two numbers it shouldn't be making them in the first plac…

You know the algorithm for arithmetic. Are you telling me you could sum any large numbers first attempt, without any working and in less than a second 100% of the time?

I don't get why the sudden fixation on time, the model is also spending a ton of compute and energy to do it

Re: Sam Altman goes before US Congress to propose licenses for building AI

#387

Post by m00x »

JumpCrisscross wrote: post #265
brenns10 wrote: post #234

Reminds me of SBF calling for crypto regulations while running FTX. Being seen as friendly to regulations is great for optics compared to being belligerently anti-regulation. You can appear responsible and benevolent, and get more opportunity to weaken regulation by controlling more of the narrative. And hey, if you get end up getting some regulatory capture making competition harder, that's a great benefit too. Open…

> Reminds me of SBF calling for crypto regulations while running FTX Scott Galloway called it the stop-me-before-I-kill-grandma defence. (Paraphrasing.) You made money making a thing. You continue to make the thing. You’re telling us how the thing will bring doom and gloom if not dealt with (conveniently implying it will change the world). And you want to staff the regulatory body you call for with the very butchers…

Except they don't make any money from their products. They're losing hundreds of millions per month.

This isn't the same at all.

Re: Sam Altman goes before US Congress to propose licenses for building AI

#388

Post by logdap »

brkebdocbdl wrote: post #350
shawabawa3 wrote: post #322

Earlier quoted context omitted.

If LLMs aren't AI nothing else is AI so far either What exactly does AI mean to you?

thanks for exemplifying the problem. intelligence is what allows one to understand phrases and then construct meaning from it. e.g. the paper is yellow. AI will need to have a concept of paper and yellow. and the to be verb. LLMs just mash samples and form a basic map of what can be throw in one bucket or another with no concept of anything or understanding. basically, AI is someone capable of minimal criticism. LLMs…

> intelligences is what allows one to understand phrases and then construct meaning from it. e.g. the paper is yellow

That doesn't clarify anything, you've ever only shuffled the confusion around, moved it to 'understand' and 'meaning'. What does it mean to understand yellow? An LLM or another person could tell you things like "Yellow? Why, that's the color of lemons" or give you a dictionary definition, but does that demonstrate 'understanding', whatever that is?

It's all a philosophical quagmire, made all the worse because for some people its a matter of faith that human minds are fundamentally different from anything soulless machines can possibly do. But these aren't important questions anyway for the same reason. Whether or not the machine 'understands' what it means for paper to be yellow, it can still perform tasks that relate to the yellowness of paper. You could ask an LLM to write a coherent poem about yellow paper and it easily can. Whether or not it 'understands' has no real relevance to practical engineering matters.

Re: Sam Altman goes before US Congress to propose licenses for building AI

#390

Post by joaogui1 »

chaxor wrote: post #120
srslack wrote: post #9

Imagine thinking that regression based function approximators are capable of anything other than fitting the data you give it. Then imagine willfully hyping up and scaring people who don't understand, and because it can predict words you take advantage of the human tendency to anthropomorphize, so it follows that it is something capable of generalized and adaptable intelligence. Shame on all of the people involved in…

What do you think about the papers showing mathematical proofs that GNNs (i.e. GATs/transformers) are dynamic programmers and therefore perform algorithmic reasoning? The fact that these systems can extrapolate well beyond their training data by learning algorithms is quite different than what has come before, and anyone stating that they "simply" predict next token is severely shortsighted. Things don't have to be '…

The paper shows the equivalence for specific networks, it doesn't say every GNN (and as such transformers) are Dynamic Programmers. Also the models are explicitly trained on that task, in a regime quite different from ChatGPT. What the paper shows and the possibility of LLMs being able to reason are pretty much completely independent from each other
Post reply on HN