cm2187

Deservedly. But the HN audience is generally on the other side of the median of the IQ distribution than Reddit’s.

But surely they see the backlash on HN and reddit. I find it odd that they don't seem to communicate at all on that.

Not really. As far as I can tell the reason Intel routed seems to be because of an explosion in one of their fabs making Xeons which reduced production, and because no one buys new desktops and servers anymore because of the cost of RAM and storage. Given intel isn't big on AI chips, they are a casualty of the AI bubble.

When I was at university, a student caught cheating was cause for being expelled.

I am recruiting college kids right now, impeccable pedigrees, elite universities. Our recruitment process involves the kids recording videos where they answer questions as a pre-screening. They are explicitly told not to use AI for the exercise. I have the Claude and ChatGPT answer to those questions on my screen while I review their responses. 40% read it pretty much verbatim. The paradox is I don't even care whether they knew the answer or not, but to me it's an integrity test.

Where are the successful alternatives?

You could have said the same thing 20 years ago about wikipedia and the internet in general. A kid in a poor school on the other side of the world has access to the same wealth of knowledge than a student in a Ivy League university. But that hasn't made universities useless, and very few of those kids in poor schools (that wouldn't have gone to a elite university) picked up the opportunity.

If anything, AI in education I think will lead to a divergence. Mediocre students will be even more mediocre as they will use it as a substitute for learning, with AI doing their homework. Good students will leverage it to learn and understand things even faster.

There are also tax considerations, even movies get shot away from Hollywood.

Fair comparison. Texas just needs to flood the valleys of its famous mountain chain and collect the melted snow with dams to match Sweden's 40% hydro electricity production.

AI will make developers infinitely more productive. You will need one or two guys to do the work that requires a whole team before. That means cheaper software.

The demand for software is huge, the number of manual processes that can be automated in any organisation, small or large, is humongous. The Jevons paradox will play at full, as software gets cheaper we will consume even more of it, and I think there will paradoxically be more software developer jobs.

But does that mean that UTC time will diverge from common time zones by more than a fixed number of hours (and minutes for a few esoteric time zones)?

Weekend hobby!

I also wonder what this guy was paid by those AI labs...

Like a selfie of me at the conference?

I was at a conference last year where one of the speakers was from National Grid. Not only these morons had in mind that electricity demand should decrease (even before AI, it was obvious that demand would go up as we are moving away from fossil fuel with both heating and transportation going electric).

But their great idea to deal with the shortage is intermittent availability of power. You won't be able to wash your clothes when you feel like, you will have to adapt to when there is spare power, like in developing countries. This is just a regression in technological progress.

I saw the same messaging in France where the current president started his mandate with a grand programme to switch off a third of nuclear power plants.

If anyone thinks adults are in charge, let alone engineers, you are greatly mistaken.

Age is an important factor. Don't underestimate that after 40-50, few people like to change their habits and learn new things (and for some that comes even earlier). Doesn't mean they aren't smart, I have seen that first hand with relatives: engineers, medical doctors, one even running for a nobel price at one point. But computers and the internet came too late for them, they either ignored them, or used them barely understanding how they work, even though they can technically understand the physics and logic of every component they are made of. You don't teach new tricks to an old dog.

AI is going to massively transform how most professions work. It will take longer than I think most people expect (I'd say 5-10 years), but there will be an unrecognisable before vs after, like with the appearance of personal computers.

Actually it looks like that changed recently, it seems to be able to render now

Seconded. OpenSCAD is a bit tedious to use from scratch but claude is pretty decent at generating the code. It can't render the script though so it tends to make silly mistakes

AI has mostly removed the difficulty of Linux configuration for beginners

But isn’t that including the cost of training, and I am not even sure if it’s training that very model?

In an interview earlier this year I remember Dario saying that the models are profitable, ie they more than pay back their inference and training over time. But because they invest in that explosive growth they have a deep negative cash burn.

And whatever markup the AI labs make, that’s on top of nvidia’s markup, micron’s markup, etc. It’s an industry where every supplier is adding a 70-80% markup!

Frontier labs subsidising? I thought they were running with 80%-ish operating margins which is part of why they cost 10 times open weight models.

Also isn't an open weight model also subsidised? Training isn't cheap and you are not paying for it.

I am not sure I understand. Surely a AI generated song isn't a "fake song"? Is it a crime to publish an AI generated song?

It seems that what he is guilty of is to generate some fake users pretending to listen to his songs, is that right? I can see how it is fraud, but how is it different from all the dark patterns used by newspapers to generate more page prints (like auto-refreshing pages)?

Irresponsible? You can safely ignore the GitHub repo if it bothers you so much. You can safely browse away.

My point is that you need to do that with 100.0% of the workforce!

Strongly disagree, and I am one of those who believe this is a bubble.

At home, Claude has been a god send to make it easy to learn using linux, and all sort of technologies. My personal learning curve has gone steep at an age where people start flattening out (40s).

At work, we use AI to automate 50-75% of our work. Better quality output, more productivity. I work in a bank. But the average plumber or electrician benefits too. In minutes they can become an expert into some tech they never touched before.

I thought crypto was bullshit, web 3.0 [corrected] was bullshit, meta-universe was bullshit. AI is bubbly, not sure how you can justify multi-trillon valuations for companies with 80% operating margin in a competitive market where you can switch to the competitor with a drop down. But the productivity gains will be massive for the economy. I think the economy will ultimately reap the reward, not so much the labs. But I may be wrong.

Active working population: north america: 195 millions, europe: 366 millions (says ChatGPT).

$6 trillion a year / 195+366m = $900 a month, from the guy flipping burgers to the hedge fund manager. Average salary is in the region of $5000 a month, pre-tax.

I love AI, spend my evening clauding. But still. Call me a sceptic on paying $1000 a month.

Low code solution have a fairly steep learning curve, and when someone got into it in anger, the complex electrical diagrams that result are even more impenetrable than spaghetti code.

In a large enterprise environment I have seen business users getting burned by vibe coding already, as they often don't even read the comments and warnings that the LLM gives them. But I think it is inevitable and should enable software to automate the millions of manual processes that are too bespoke to pay an IT team to automate.

For a solo entrepreneur, it should enable to get so much done before having to hire the first employee. Hence less risk to go bust, easier to take risks, pure value added.

Can LLM access FoxPro programmatically, or is the file format easily parsable? Most of FoxPro apps are likely fairly simple CRUD apps. It would be trivial for a modern AI to transcode it to a web based client server solution.

Exactly, or an access database. This is to enable end users who live in a fully locked down environment (enterprise setup) and can't install anything to use database like features without having to wait 6 months for IT to lift a finger.

But if it starts being a system of records / authoritative system, there needs to be a plan to decommission it.

The only difference I see on the facts vs what I read so far is that OpenAI knew of the hack and decided not to intervene. I haven't read the report but I find it hard to believe that OpenAI deliberately let its agents hack a third party company during a training run. It would certainly attract a criminal liability, which would be surprising to admit in writing.