Most of the developed world has lower homicide rates than the US, and has had for a long time.

Lol. You're just determined the world is ending hey?

Right. Those are some of the hot topics in lunar science. So if someone lands a couple of rovers there with a bunch of cameras, penetrating radar, spectrometers, etc. finds water, and analyzes it, they'll be at the forefront of lunar science.

If they find water it will certainly put them at the forefront of moon science.

By the way, those articles seem fine. The first two are well established history at this point.

The person you answered to has nothing, especially not any understanding of what AI research took place in the last decades. Hence they just try to confuse the issue and introduce something completely irrelevant as the new main topic. Obvious and unsophisticated.

The reality is that the LLM fans are essentially cultists and that they have completely lost sight of all the limitations we find in LLMs. Sure, LLMs are not useless. They allow better search (with sometimes really bad hallucinations and misinformation), they can summarize things in a way that helps experts (non-experts may be grossly misled though). What they cannot do competently is synthesize anything. Text-slop, code-slop, code that on repeated changes re-introduces regressions (which competent coders will not do), and even art- and music-slop. So, to better deal with the mountain of data we have today, they are somewhat helpful to actual domain experts, but that is it.

But the LLM fans think that the synthesis aspect is pretty good as well. It is not, at least to anybody that is a domain expert. My guess would be these people cannot perform on professional level either and hence think that slop is just fine. For example, the only mostly adequate code that LLMs can write is attack code from patches or patch-notes, because that is use-once, throw-away code. For anything else, as long as it is not beginner's level, an actual expert has to request in detail what is needed and make sure it is all there. At realistic token prices, that seems to be a lot more expensive than to have an expert write the code and that does not include the cost of low quality aspects that are not immediately obvious, like LLM code being long, hard to read and hard to review.

There is no reason to expect drastic cost reductions either. The current LLM business models need growth-rates they cannot get in the real world. Most of the current providers will crash and vanish. The rest will drastically reduce what they offer or charge prices that make most uses for synthesis non-viable.

Add that we will very likely find other serious drawbacks of having LLMs write things. This tech is on "demo" level currently and the demo was scaled up beyond all reason. Maybe it will deliver something reasonable in the synthesis use-case in 20 years, but that is not assured at all and not only because the inevitable crash long before.

Just because it is sold cheap doesn't mean it isn't valuable. It may just be that people who value it don't know it exists, or it isn't available due to international shipping, or they won't start looking for years to come.

That's why I get scanned books onto the Internet Archive. Maybe nobody will care, but maybe someone will 10 or 100 years from now.

Competition is usually good. YouTube has been rapidly enshittifying, and there isn't really any alternative.

You can have both physical controls and a big screen. I like having a big screen for Android Auto, but wouldn't give up the buttons for other functions either.

Some of that stuff is good. Autopilot systems that actually work are great for long distance driving. You need some compute power for Android Auto/Carplay, and having a built in sat nav can be useful too. Remote management of air conditioning and charging are very nice to have.

Of course BMW's ads and other bullshit would be enough for me to return the car. My car has physical buttons for key functions too.

If they do find ice there, it's going to be huge. And an issue for the US, because everyone wants to go look there, but if the Chinese find it first it will be politically difficult to justify following their lead.

Unfortunately cooperation is too much to hope for, but it looks likely that China will get boots on the ground there before 2030 either way.

Indeed. The main theory I have is this is an experiment how many utterly dumb people and how many complete assholes you can mix in without the whole thing becoming dysfunctional. I am not sure whether this is a success or a failure so far. I am tending towards "failure".

... without any or any good justification, this could work. Or not, if Bilibili turns out to be worse. In any case, YT has overstayed its welcome a while ago.

Naa. People are generally pretty dumb without social media. All TikTok will do is stop them from more failure using the pathetic excise for rationality the own.

Indeed. But the morons praying to the tech-god are not rational. They do not see what the actual reality is. And the are aggressively attacking anybody that calls out their delusion. Nothing new. Still utterly pathetic.