dauertewigkeit

The Rest is History is good, depending on the topic. Both guests have a bit of bias which you have to sort of take into account, not that different from The Rest is Politics. Mishal Husain has a new podcast on Bloomberg TV which so far was excellent. Also from Bloomberg TV, Big Take is often interesting. I still enjoy Lex Fridman, again depending on the guest. Dwarkesh Patel same shit as Lex, but he pretends he knows something about AI.

Western politics is all about constructing these narratives that hide the hypocrisy and self-serving nature of the dominant political factions. You can see it everywhere, but this is one clear example of it.

It's a very interesting read, but a lot is not clear.

How does the seller get these desktops directly from NVIDIA?

And if the seller's business is custom made desktop boxes, why didn't he just fit the two H100s into a better desktop box?

The farmers and the sawmill operators are easily explained. I also listen to Bloomberg, lol. Those people are rich and most of their wealth is now diversified away from their businesses, and while they would rather keep those businesses alive because it's part of their family identity, they care more about reducing their taxes on their overall net worth.

US just has more to offer.

It's a small country, relatively speaking. Rather dull cities, again relatively speaking. Rural land is hard to come by and expensive. Not a lot of sunshine hours either. Not English speaking, not an immigrant culture, and quite an insular society so if you're not born there it kinda sucks. The cities punch way above their weight, but in total the tech job market is still tiny compared to the US. If you like being outdoors, Switzerland has one landscape, pretty much. It's heaven for rich people, but a very specific kind of heaven.

I'm not disputing the facts. I'm just pointing out that it's the norm.

I'm not sure what you're implying though. I don't think he is being platformed by current mainstream institutions if that's what you're saying.

It approximately is, imho. Wealth follows a power law distribution. People put the dividing line at different points, but it doesn't matter so much. The elite are a tiny fraction. The middle class are also a relatively small faction of the population and for the most part, the middle class tend to be lumped in with the elite, because they tend to be in complementary political factions.

Now I know that in the US, people group everyone with a job in the middle class, but that's just semantics.

I don't think that's abnormal. That's the norm for political leaders in western countries. There are very, very, few people that rise to leadership positions from a purely working class background. Even Jeremy Corbyn grew up middle class. It's almost tautological that leaders are going to be above average in some respect and this talent will be recognized early and the way it works in western countries, the elite institutions try to recruit all the talented folk from non-elite backgrounds into their ranks.

I think it's overall a good thing that not all people from elite backgrounds with above average IQ/skills end up being purely upper class aligned.

I've been saying this since before Yanis was even a Greek MP. It's just so damn ironic that capitalism and free markets ended up building these huge corporations which are essentially planned economies at nation state scale.

Correct me if I'm wrong but LeCun is focused on learning from video, whereas Fei-Fei Li is doing robotic simulations. Also I think Fei-Fei Li's approach is still using transformers and not buying into JEPA.

It's a middle school essay that is trying to score points based on the number of metaphors used. Very unappealing and I wouldn't call it technical.

EDIT: For all the people saying the writing is inspired by math/cs, that's not at all true. That's not how technical writing is done. This guy is just a poser.

CNNs and Transformers are both really simple and intuitive so I don't think there is any stroke of genius in how they were devised.

Their success is due to datasets and the tooling that allowed models to be trained on large amounts of data, sufficiently fast using GPU clusters.

Tesla's problems with their multi camera non-Lidar system is precisely because they don't have any spacial cognition.

Sutton: Reinforcement Learning

LeCun: Energy Based Self-Supervised Learning

Chollet: Program Synthesis

Fei-Fei: ???

Are there any others with hot takes on the future architectures and techniques needed for of A-not-quite-G-I?

Millennials are really great parents and the result of that is that the kids are well rounded and less deficient. That results in conformity because the history nerd, also goes to the gym and the gym bro also strives to do well at school.

better connectivity -> people finding better friendship matches -> groups are more homogenous -> more polarization

Why is that bad? When the topic is abortion, not being born is considered a good thing for the child, whose life prospects aren't so good on account of the economic conditions of his mother.

If you are a cleric, you are not supposed to be involved in any romantic partnership and sex outside of marriage is not allowed. As such it makes no difference if you are straight, gay or anything in between.

At first I thought it was a passport bro on vacation, but it turns out it was a love affair (cheating). Everything about this is disgusting and whoever pushes this shit as some kind of romance needs a kick in the butt.

I said nothing about belonging to different economic classes.

My point is simply that if you are 1 of 50,000 in a meat processing plant, you do not really have any way of competing with your fellow workers. You might try to work harder and faster, but then you end up raising the bar for all, and you now have to maintain the new pace. And once the rest catch up with your pace, the pay for all will be lowered again.

In tech, every job is slightly different, and there is a real opportunity to meaningful differentiate yourself from the rest and compete in a much more dynamic job market.

Also, who are the ones hurting currently? The juniors or the seniors? Seniors are mostly doing fine. Juniors are the ones hurting. And juniors would have an even harder time, if you transitioned to a unionized system, because entry requirements would be raised significantly to account for the fact that you cannot fire people so easily.

Good point. I am not familiar with Hollywood to know what their job market is like.

EDIT: another commented mentioned that NBA players are also unionized. I think there is a second element to it, which has to do with how monopolized the employer market is.

What makes solidarity between workers possible is the homogeneity of their labour. This condition is not present for tech workers. Labour market conditions are completely different from those of say, a meat processing plant or a factory assembly line.

I experimented with asking ChatGPT to give me python scripts to run directly as FreeCAD macros and it does OK. There is potential there.

I find the marketing interesting considering that this product already exists in other continents ... and it is NOT deployed in factories. It is deployed in office settings. If this our future? Lots of video evidence: https://www.youtube.com/results?search_query=+Using+AI+to+mo... Expect dishonest marketing by those aspiring to build these surveillance and anti-freedom systems in our western countries.

Isn't conscientiousness a biologically driven psychological trait? I always assumed that to be the case. I don't think you can train conscientiousness or any other psychological trait for that matter. But I am no expert and would be delighted to be convinced otherwise.

I agree with you that Sabine often talks about matters far outside of her expertise, but as somebody with a foot in academia, I would bet that a very large number of academics have at least one academic research direction in mind that they would categorize as a "scam to get funding".

The thinking in Europe when it comes to AI, is "AI for industry", or "AI for science". Like they think of AI as providing a supporting role in industry or science, but they do not think of AI as a goal in itself. This thinking is analogous to the way EU leadership treated "tech" for the past three decades and look how well that turned out for us...

If we had good leadership, these funds would be used to setup 2 or 3 competing AI labs. Forget all that "AI for industry" and "AI for science" nonsense and just catch up to the US and China, and have something competitive, for once.

Do you access to a computer? Maybe at a public library? Or on a university campus? You can recover your files with ADB.

It ironically seems like a very similiar market to internet search. There was no moat there either, other than the capital needed to bankroll a better search engine. A lot of these AI companies will eventually fail (not because their models will be significantly worse but because of failure to commercialize), the market will consolidate with only a couple of players (maybe two in the US, one in China and maybe one in Russia). And once that happens the idea of raising enough capital and building a competitive AI company will seem impossible. Exactly like what transpired with internet search after Google won most of the market.