HardCodedBias

"The real question is whether the boom is, economically, a mistake."

The answer to this is two part:

1. Have we seen an increase in capability over the last couple of years? The answer here is clearly yes.

2. Do we think that this increase will continue? This is unknown. It seems so, but we don't know and these firms are clearly betting that it will.

1a. Do we think that with existing capability that there is tremendous latent demand? If so the buildout is still rational if progress stops.

Gemini 3 is in preview and it is pretty clearly behind multiple iterations for OAI and Ant.

It may even be behind open source at this point.

Seem telling.

Gemini is used in agentic setups by literally dozens of people. Dozens!

Probably true.

Not sure about “by far”. But probably best.

Of course it makes sense.

The regulatory framework is getting more and more difficult for data centers.

The options are move to countries with less of an uphill regulatory burden (UAE?), but this comes with other issues.

Space it is.

Government employees have power and they are flexing the power.

C'est la vie!

Decades?

I mean, yes, the probability of having that level of tech in decades is quite high.

But the technology is moving very fast right now. It sounds crazy, but I think that there is a 50% chance of having ready player one level technology before 2030.

It's absolutely possible it will take more time to become economical.

Once their vision for "grab and go" vanished due to technological infeasibility [1] the entire premise for the stores vanished as well.

I suspect that they wanted to take a hail marry to see if somehow it was possible to get much greater efficiency compared to standard grocers, and it looks like that failed.

[1] it may come back. The technology is rapidly improving but they have bigger fish to fry ATM.

when I hear of industrial uses of phosphorus my ears prick up since phosphorus is a key limiting factor for life.

A world where this actually became industrially very successful combined with a lack of recycling could potentially add large new sink for phosphorus.

In general, be careful when creating a process which locks meaningful amount of phosphorus out of the biosphere.

I think you're right, but the issue goes deeper. If the productivity gains are real, the incentive to bypass security becomes overwhelming. We are going to see a massive conflict where compliance tries to clamp down, but eventually loses to 'getting work done.'

Even if critics are right that these models are inherently insecure, the market will likely settle for 'optically patched.' If the efficiency gains are there, companies will just accept the residual risk.

They will simply be more expensive.

More likely they will not use leading the leading edge fab process, which TBH is fine for the vast majority.

If it were only 20B then Apple would jump at the chance.

As would almost innumerable others.

Nvidia direct silicon revenue is higher.

Also Nvidia's margins are higher which means that they will be willing to pay a higher unit price.

This seems like an open and closed case from TSMC's side.

He was a brilliant observer and reporter on the behaviors of humanity.

He will be missed.

Think of how this opened up EM:

https://ddcolrs.wordpress.com/2018/01/17/maxwells-equations-...

OpenAI has #5 traffic levels globally. Their product-market fit is undeniable. The question is monetization.

Their cost to serve each request is roughly 3 orders of magnitude higher than conventional web sites.

While it is clear people see value in the product, we only know they see value at today’s subsidized prices. It is possible that inference prices will continue their rapid decline. Or it is possible that OAI will need to raise prices and consumers will be willing to pay more for the value.

It seems pretty clear.

Never forget the continual parade of "experts" when the chip ban was being promoted who all said in unison:

"China is 20 years behind"

Horribly dishonest. But they all talked in lockstep: lawmakers + "experts".

"My point is that you can actually reduce it all to dollars."

I'm not sure. A couple of points:

1) The regulatory landscape is enormous. It is unknown from which angle regulators will "slow you down."

2) As I mentioned the regulatory frameworks in this area are evolving very quickly. It is unknown what the regulations will be in 1, 2, 5 years and how that will impact your business.

I suspect that this orbital data centers isn't entirely about dollars (No doubt dollars are important).

I suspect it is about the regulatory environment. The regulatory environment on data centers is moving quickly. Data centers used to be considered a small portion of the economy and thus benign and not worth extorting/controlling. This seems to be changing, rapidly.

Given that data centers only exchange information with their consumers they are a natural candidate for using orbit as a way to escape regulators.

Further, people are likely betting that regulators will take considerable time to adjust since space is multinational.

Huge fan that Gemini-3 prompted OAI to ship this.

Competition works!

GDPval seems particularly strong.

I wonder why they held this back.

1) Maybe this is uneconomical ?

2) Did the safety somehow hold back the company ?

looking forward to the internet trying this and posting their results over the next week or two.

COMPETITION!

Unless:

1) LLM advances stop 2) The Chinese companies release open source/weight models which are as good or better than the West 3) Apple somehow turns it around with AI

Apple is done for.

AI is going to be central to the next generation of phones and the next form factor.

Their complete failure on AI has been ... shocking. Not sure if they don't have the data to train a leading edge model or if they have some kind of personele issue, it has just been shocking to see their lack of progress.

No doubt Apple has rested on their laurels for a long time. I just would not have expected this.

"There are four lights"

And the AI has been RLed for tens of thousands of years not just a few days.

MTG does better than Nancy?

https://www.quiverquant.com/congresstrading/politician/Marjo...

https://www.quiverquant.com/congresstrading/politician/Nancy...

Sure doesn't look like it.

It looks like she buys and holds, unlikely nancy. And it looks like she bought some AMD which had great returns (oh how I wish I bought it) and held it.

Looks like a false claim to me.

This is the system working.

Competition is all you need.

Measuring the value is very difficult. However there are proxies (of varying quality) which are measured, and they are showing that AI code is clearly better than copy-pasted code (which used to be the #1 source of lines of code) and at least as "good" (again, I can't get into the metrics) as human code.

Hopefully one of the major companies will release a comprehensive report to the public, but they seem to guard these metrics.

Two claims here:

1) LLMs have failed to live up to the hype.

Maybe. Depends upon's who's hype. But I think it is fine to say that we don't have AGI today (however that is defined) and that some people hyped that up.

2) LLMs haven't failed outright

I think that this is a vast understatement.

LLMs have been a wild success. At big tech over 40% of checked in code is LLM generated. At smaller companies the proportion is larger. ChatGPT has over 800 million weekly active users.

Students throughout the world, and especially in the developed world are using "AI" at 85-90% (from some surveys).

Between 40% of professionals and 90% (depending upon survey and profession) are using "AI".

This is 3 years after the launch of ChatGPT (and the capabilities of chatGPT 3.5 were so limited compared to today that it is a shame that they get bundled together in our discussions). I would say instead of "failed outright" that they are the most successful consumer product of all time (so far).

I thought Ilya said we have more companies than we have ideas. He also noted that our current are resulting in models which are very good at benchmarks but have some problems with generalization (and gave a theory as to why).

But I don't recall him actually saying that the current ideas won't lead to AGI.

Well of course it is not enough. This is well known.

"Success" is hard to nail down. Is it academic success? SES succes? Job performance? It's all over the map.

However we know that :

IQ is correlated with the above somewhere between 0.2 and 0.5, with many definitions and studies being near the top of that range.

Conscientiousness is correlated with the above somewhere between 0.2 and 0.3, with many definitions and studies being near the top of that range.

Low neuroticism is correlated with the above somewhere between 0.1 and 0.2, with many definitions and studies being near the top of that range.

And there are other "personality" metrics that have been studied. It is very easy for someone who has an exceptional IQ to be sert back by neuroticism for example and exceptional IQ is near useless if the person does not have the conscientiousness to follow through on tasks, these people will likely be exceptional a "shallow" tasks.

I think this is well trodden ground.

@complaintvc on X has been doing amazing work in this area.

The EU, especially the EU post 2008, seems to be infatuated with regulation it has likely bitten them with their lackluster GDP growth and their very lackluster AI developments.

I suspect that this is too little too late, and more importantly I highly doubt it signals a shift in the biases/incentives of the EU regulators. The second the scrutiny is off of them they will go back to their ways. It is their nature.

(I look forward to the loss of karma. I hope that the link to @complaintvc at least makes a few people chuckle).

"Hackers should understand governments are complex, dynamic and occasionally chaotic systems"

No. Hackers should understand that government is force. This is the definition of government.

And force is the antithesis of the hacker ethos.