Mightn't be so bad if all the value wasn't being captured by trillion dollar corps, despite them benefiting off the labor and creativity of countless non consenting individuals

I’m not sure people are “wishing for the death” in the literal sense.

We just need to apply the antitrust law and break these trillion-dollar corps into smaller competing corps.

There’s no reason for a single company to provide your email, browser, search, operating system, entertainment, AI, etc.

How many of those jobs do you expect them to retain as they build increasingly capable models?

How much of the profit will be redistributed to all the artists, programmers, scientists, etc. whose works are the foundation of their product?

The poorest 50% of global citizens hold the same combined wealth as the top 10 individuals.

How are massive and increasingly concentrated corporate profits going to help them? (Or anyone else for that matter).

>People wishing for the death of the mega corps, are directly wishing for the obliteration of millions of great jobs.

Yeah, I'll take the obliteration of a few million overpaid American jobs over the destruction of hundreds of millions worldwide, the social upheaval that goes with it, the destroyed families and the direct and indirect deaths. Fuck your promo packet and fuck your cozy desk job.

well they won't give me one of those great jobs

Cry me a river

I mean.. There are degrees of control and agency right? I really don't understand the impetus to pretend AI is just doing exactly what its told.

There is a lack of due diligence and effort to restrict and properly configure AI to operate within the bounds of the law.

sure, but when you read beyond the headline in these events the story is never that the AI went rogue and disobeyed instructions. it's always that somebody gave it instructions to go make a mess, and then is surprised when the AI makes a mess.

a human decided that control and agency surface, then clicked go

a human is always behind it and ultimately responsible

To avoid sarcasm, isn't the term "artificial intelligence" part of the answer to your puzzlement? If people object to "stochastic parrot", and want some more precisely descriptive term like "artificial intelligence", then why do we experience surprise when people act as if that system is intelligent?

My puzzlement is that people object to attributing agency to things that are clearly "agent-like". That doesn't mean the developers and companies building it get a free pass.

Presumably this is only a bright point if you dont think its false hope?

This seems highly optimistic to me.

What's to say current AI won't be highly power-concentrating by default?

The best models are owned by a few companies, and displacement of knowledge-workers mainly seems to benefit the capital class.

On-device / edge computing makes sense in a few very limited scenarios. And economically, price or watt/token (or watt/task completed) might always be better in large data centers.

No reason to think we are on a trajectory towards broad empowerment right now

> The best models are owned by a few companies

Whenever I see 'best model,' 'most advanced model,' etc, it just reads to me as 'roundest ball'. The new model is the most precisely round ball ever, models next year are going to be even rounder, etc.

The difference in utility between the latest, most round ball and last year's frontier balls (which are now freely available to the public) is pretty debatable, imo.

Making balls ever rounder clearly has diminishing returns, as there's a limit to roundness.

Is there a similar limit to intelligence though?

Yes, absolutely. Even human intelligence stops being useful at a certain point and becomes 'enough, provided the human is motivated and tenacious'. Intention is a better metric, but it's a serious AI weak point if not actively an achilles' heel. 'Warios' come to mind.

Intelligence in humans probably stops being useful for the individual at some level, indeed. But that might be due to being in an environment that's unable to feed that intelligence and execute on the ideas in generates.

“I am, somehow, less interested in the weight and convolutions of Einstein’s brain than in the near certainty that people of equal talent have lived and died in cotton fields and sweatshops.”

― Stephen Jay Gould

Interesting. Any evidence that intelligence stops being useful? Why do you believe this is this a bounded resource.

Numerous studies have shown that the iq/income correlation breaks down at around 130 IQ points in humans.

> The best models are owned by a few companies

In terms of cost per task, the open weight Chinese models are winning by a long shot. So it depends on what you mean by the best model.

Don't you think open weight Chinese models will be the first thing they start banning once the regulation of this industry starts to kick in?

Yes, but if you read up two messages in this thread they’re discussing whether AI is power concentrating by default. If you concentrate power through regulation, that’s not “by default”. That’s just corruption in government resulting in a concentration of power.

Are you sure about that?

when I checked a few weeks ago I did not really find anything competing per value with a 20x Codex subscription.

In terms of API costs they're about right, 10% of weekly usage on a 5x sub is roughly $40 (for Astra). Chinese models cost roughly half as much (and have fewer refusals).

Subs have insane value because large companies cannot use the subscription model. They are loss leaders and are often used for passion projects & startups (where the juicy data is).

Ah and OAI stopped offering the 20x sub as of yesterday.

This is only correct if you ignore subscriptions. I ran the numbers on what you get for your subscriptions last week: https://blog.brokk.ai/a-coding-subscription-tier-list/

That’s only one metric. It doesn’t matter how cheap it is, if it continually fails to solve a problem.

Yes they’re getting better. No, I don’t think this will be a panacea. If only for the fact that this is China (more specifically the CCP) after all - they’ve got their own plan, and altruism is not part of it.

TBH it doesn't need to completely solve a problem.

I can use GLM to do in 2 hours what would have previously taken me a week.[1] Switch to Astra or Fable might take that down to 1 hour, maybe.

The difference is not so big when you look at it that way.

==================

[1] Actually, no, but let's pretend for the sake of argument that SOTA models can 10x - 40x your productivity.

Have you actually used the latest Chinese models? Because, in my experience, they're not just cheaper, they're actually better in many cases. In one recent experiment I did a few days ago on a task that I need in production at scale at my company, Qwen 3.8 and GLM 5.3 Flash were not just cheaper, the results were significantly higher quality than GPT 5.6 Sol.

Even if China doesn't continue to release this stuff for free, I think somebody will. Eventually, maybe Europe or maybe a smaller country that picks up this knowledge will. Or maybe just some random philanthropic billionaire.

Yes, I use the messing Chinese models extensively. Like you, I very much appreciate them and sometimes prefer them.

But there is a bar of complexity at which they fail, and repeated invocations typically doesn’t make much progress. This is a small subset of most work, but it still exists. And yes you can help it along, but in those cases I’d typically break out the big guns.

> Qwen 3.8 and GLM 5.3 Flash were not just cheaper, the results were significantly higher quality

How can they be higher quality than models they were distilled from?

> How can they be higher quality than models they were distilled from?

The word "distillation" is not specific to AIs, has been in use for 100s of years, and does not mean the same thing as "dilution".

It means "getting a more concentrated form of the original product".

Very shallow answer to a valid question. The vast majority of people know that word btw, that's not some revelation.

Fair. My point is not that all of Marxism is correct (or even most of it) but rather that the specific pattern he and other theorists in that school of thought observed when developing the theory of historical materialism seems to strongly match a lot of what's playing out in AI.

AI, doesn't feel all that different to the mechanical loom that was the impetus behind a lot of Marx's early critiques of industrial technology.

You've got a really cool invention that replaces what was previously a skilled artisan trade. The wealthy then use this invention to excoriate the artisans and render the loom operators as replaceable commodities.

Dario is essentially writing a blog post about why only he (and a few other approved elite companies) should be trusted to own the loom and then make it available for everyone else to labor at. Like the capitalists of Marx's era, he genuinely believes it is better for everyone to have this arrangement and that it is natural people like him should sit as benevolent gods at the top of the pyramid.

It's fascinating to me to see the same underlying mechanisms that gave birth to some of the greatest failed political and social experiments of the 20th century bubbling up again so clearly with AI today.

What does this even mean. There is strong evidence of LLMs doing in context learning.

Some of the linear RNN layers in recent models are provably doing SGD in hidden space during inference

By in context learning do you mean latent space?

https://transformer-circuits.pub/2021/framework/index.html

> What does this even mean. There is strong evidence of LLMs doing in context learning.

1. Is this learning persistent?

2. Do they verify these new lessons against core principles?

3. Do they and protect themselves/ignore requests if these new lessons contradict those core principles?

Humans do that from the time they're 3 years old (not that well, but they do do it).

Yes. Yes. Yes.

In my experience all those claims are false.

So the next step is to ask for evidence and ideally independent and peer reviewed research.

You know they can take notes, right?

And ICL dates all the way back to 2020, at least: https://arxiv.org/abs/2005.14165

That's training during training. They can't learn afterwards.

do you have a reference where that claim is demonstrated?

What unique differentiator could musk possibly offer though?

Is there really a high demand for racist chatbots with no safety controls?

And even if there is, will governments even allow it to keep operating?

I find the presentation of this as "factoring training" very confusing.

Factoring is something we do to distributions, which we then parameterize with our neural net.

An autoregressive model pertains to a certain factorisation of a joint distribution. And if we choose this factorisation it may have consequences for our training and sampling etc.

But to say "most models on factor generation".. I dont quite get this.

This would be better work if it didnt claim it was a new pretraining axis also. Its another way of scaling compute right?

If it were a new axis, IWAE would have an equal claim to it (as others have pointed out).

I immediately thought of IWAE as well.