Panoramix

How do we know the formalization itself doesn't have any errors? AFAIK it's very far from trivial to formalize something.

This is lovely

It's explained in the video. They target <1 second.

Do you have a citation for that?

Show me a video from 50 years ago of a car doing this: https://www.youtube.com/watch?v=7p5LxFdc-Qg

It's just plain impressive, even if it's not practical.

I guess nobody clicks on the links anymore? From the OP: "Our multi-mode fan and ground sealing system acts as a downforce generator, pulling air out from beneath the vehicle. This provides enough suction to hold the vehicle upside down, even when stationary. Powered fans and active wings maximize traction and cornering speeds in all conditions, achieving massive downforce at zero speed. A next-gen steer-by-wire system offers incredible feedback in various road conditions. Activate Track Mode to lower the car and transform the suspension. High downforce or low drag modes can then be used as needed on the track."

Sure. But no European company is ordering more than a few DUVs at a time

The logic is backwards. The research was already there, so ASML licensed it. If the research wasn't there they would have developed it, and/or find another way to make it work using a somewhat different tech. This is but one of the dozen large technical issues that needed to be solved en route to EUV.

Also, if you start now you would not compete with TSMC in the coming 20-30 years, so the energy requirements won't be that high off the bat

And about 300-400 DUV machines a year

Your point is largely addressed in the article, did you try reading it? "In many fields and activities, years of training have traditionally served not only to produce a final answer or product, but also to develop understanding and the ability to formulate new questions and ideas. However, building on a vast body of previous human work, AI systems are becoming increasingly capable of producing the results of such work directly, and these goals cease to align."

The point is that these proofs are largely useless without the insights. The value of a proof is largely in the travel, not so much in the destination.

Yes, that would be par course with OpenAI behavior

I mean it kind of makes sense that the people that caused the problem don't even see the problem. That's an explanation onto itself.

That doesn't work.

You're right, and the load-bearing part of the argument is not what you think it is. Two ambiguities worth resolving before moving on: whether what you wrote also applies to ChatGPT, and whether you have custom instructions set up. Failure mode worth flagging explicitly: I didn't read TFA.

(I'm becoming allergic to how these things write).

For those that don't mind a lot of rootkit and embedded spyware you mean

Not me but I know plenty of people that would volunteer

A bunch of ideas doesn't make a book. Actual writing takes place, the length of each sentence, the specific choice of words, the pace, the choices, the editing. It's literally an art. Nobody wants to read what ChatGPT wrote based on your prompts.

Of course you can use it to quickly try out stuff, for some editorial tasks and such. But unless you learn the craft of writing I don't think others will want to read any of this stuff.

The point is that now it's dramatically worse by volume

I don't know about that, a lot of those bloated enterprise solutions are becoming obsolete fast, and integrations becoming a moot point as these models advance.

What does that have to do with ChatGPT?

Sushi chefs spend years learning the correct feel of the fish - when it's warm enough, when it's slimy. Japanese are taken aback when they are forced to wear gloves for "safety", which at least in that case is entirely counter productive.

All the things you wrote, but also quality. Quality is a giant issue. Garbage usability. Excel is the only Microsoft product I thoroughly respect. The amount of things you can't do in PowerPoint for example is mind blowing.

It's about the thermal conductivity.

Helium has 150mW/mK vs Argon ~18mW/mK so you can't replace it.

The only alternatives to Helium are Neon, which is 3x worse and much more expensive, and hydrogen. However, hydrogen is flammable so it's a very bad idea to use it in a fab which has extremely poisonous gases and needs a cleanroom environment. A fire would ruin your whole factory and kill your engineers.

I just tried 5.1 and got the exact same output as for 5.2 (actually I got slightly less info with 5.1)

Think of things like your preferred units (meters, kg, cups, tablespoons, milliliters). Or, do not suggest recipes with x ingredient. Language preferences. Etc etc etc.

There's several options, and in the future there will be even more. Labs like Mistral, Sakana AI already have good products; they will only get better.

More reasons to go with the competition

Nice overview. Some of the descriptions are quite thin on details, like "new model by x", or "latest model by y". Well of course it was new at the time but that doesn't really add information.

Yes but how do you do that? that magical third electrode sounds harder to make than the original problem.

Edit: I think I get it now, it's a chemical reaction. By applying a voltage with some polarity to the 3rd electrode you can run the reaction in reverse. Still very hard to achieve because you have to make sure the reactions happen at the same rate with the same efficiency, which is far from trivial. This must be a very high end sensor for all this effort to make sense.