zhangjunphy

I think if we can have a good enough simulation of reality, and a fast one. Something like an accelerable minecraft with real world physics. Then this idea might actually work. But the hard reality we currenly could generate efficiently and feed into LLMs usually has a narrow scope. It feels liking teaching only textbook math to a kid for several years but nothing else. The LLM mostly overoptimize in these very specific fields, but the overall performance might even be worse.

True, and the successful ones usually require an external source of information. For AlphaGo, it is the simple algorithm which decide who is the winner of a game of Go. For GAN, it is the images labled by human. In these scenarios, the critic is the medium which transforms external information into gradient which optimized the actor, but not the direct source of that information.

I also hope we have something like this. But sadly, this is not going to work. The reason is this line from the article, which is so much harder that it looks:

> and a critic model filters the results for genuinely valuable ideas.

In fact, people have tryied this idea. And if you use a LLM or anything similar as the critic, the performance of the model actually degrades in this process. As the LLM tries too hard to satisfy the critic, and the critic itself is far from a good reasoner.

So the reason that we don't hear too much about this idea is not that nobody tried it. But that they tried, and it didn't work, and people are reluctant to publish about something which does not work.

It is a counter argument against that we can build a perfect simulator of our universe, but not a good one against that our universe is simulated.

In fact if I were to build a simulator, I most likely have to design a mechanism to prevent its residents from observing beyond a certain micro scale due to limited cpu/mem resources and laziness to implement all the details. Tiny black hole is a good mechanism to reduce resource consumption when simulating a fixed volume of this universe. Imagine living in the world of Minecraft, the minimal unit is a block. Trying to look inside of it yields nothing. All physically meaningful characteristics are described by its surface. In our universe this looks very much like a blackhole.

The market has changed drastically since some Chinese factories start mass producing synthetic diamonds around 2019.

Currently in the end market the price is $200~300 for a 1CT synthesized one, and the quality are simply better all around. The only way to distinguish natural ones is to track from the start where each is mined and processed, which is a bit absurd.

Many of these factories were associated with drill head industry and the like until they found the jewelry market is much more profitable and got the trick to mass-produce them. If they are willing to compete a bit the price might even go lower.

QFT is too successful that experimentalists stuck for half a century with no meaningful surprise. But theoreticians got to maintain a publication streak if trying to stay in academy. Thus string theory becomes the perfect field. Most ones doing it seem to know it is a goose chase but no one dare to admit it publicly as your colleagues stripped of funding would tear you to pieces.

I feel we really should sort these papers into a category "physics in imaginary universe", so other people do not get confused.

I guess it really depends on what you are working on.

If you are writing some non-trivial algorithms or working on some projects which requires delicate handling of things, then Copilot is most likely going to mess up.

But if you are working on many of those frontend code or backend CRUDs which are usually quite repetitive. Then Copilot could be helpful.

It is really interesting to read the comments in this thread.

From an outsider's perspective, US has been the most prosperous country in this planet. People generally have good chances to move up the class ladder. And there are a lot of active enterprises. But I had the strong feeling that this stop to be the case in recent decades.

The reason, as I see it, is the natural tendency of capitalism, people seeking to maximize their individual capital. In the past century, with the pressure of the cold war, this tendency is balanced by the government, as US needed its working class to actually make innovations and make products to gain the advantage in the competition with the Soviet Union. In fact, the working class got their best treatment during the cold war, and then it worsens in recent decades.

Now with the Soviet Union dead, and China is still too young to be an actual competitor. There is no obstacle for the capitalists to play their games. It should be evident for most Americans that the price of properties are inflating while the increase of their salary slowed. The government also got eroded, most of the people there are either capitalists or have good relationship with them. What will be their incentives to change the rules?

This is not just the problem with US, it happens also in China. And I don't see a solution to it. The lower classes are controlled tightly by the media, which is controlled by the capitalists. And the middle class is thin and divided. In fact a lot of the middle class try very hard to climb the social ladder so they can be one of the upper class. They don't want the game changed, just "modified" to benefit themselves a bit.

Revenues are important. The users will not notice unless something happens. And when something happens they forget fast.

than to be not free and starve