Walking on two legs is not a smart mode of locomotion for a machine.

If you do not find fun as an adult, you are doing it wrong. Here is a hint: Regressing yourself to a mental child is not the way to do it.

Fun part: you can tell properly trained AI to give generated piece "a sense of the personality and mood of the specific performer". Today, 4/5 times it will be successful in doing so, provided it had enough of performer's works to learn from.

At this point, there are two kinds of successful artists.

Those that tell you that they use AI to assist with their craft, and those that lie about it.

Because the amount of extra work that needs to be done to without AI is ridiculous at this point. Add to this the part where AI provides several unique services that humans cannot deliver in time frames that aren't thousands to tens of thousands of times longer, allowing for rapid iteration on concepts that would be impossible otherwise, and today you no longer can be successful without it.

The only difference today is how much of it is used, for what parts, and what specific model.

Nope. But idiots incapable of learning, like you, are still expecting the golden age tomorrow when it has been announced to manifest very soon for 30 years now.

QCs are not going to happen. Deal with it.

I agree. But what this means is the death of low and medium expertise levels. People that get high expertise have always found ways to do it. They are just pretty rare (maybe 10% of all CS/IT people), and that makes them expensive. On the plus-side, we have been producing way too much (mostly crappy) software anyways and we have used way too high-maintenance server and software systems in addition. The 10% are quite enough to keep things running and maintained once all the cruft and crap is removed from service. Obviously, AI will not play a role in that, because it cannot and will never be able to perform on that level.

What a stupid comment. Did you even read what the story was about?

That is the wrong approach. To firewall these, you need to know the type and nature of the backdoors in there. But once you know that, you reject the product.

Also, a firewall does nothing for SMS unless you do SS7 firewalling, which is very, very hard to do and requires expensive equipment.

Now if all these idiots claiming Chinese backdoors would either shut up or deliver evidence.

That is not the point. The point is that new code gets the minor security benefits of Rust and the major benefits that all the mediocre and bad coders will be unable to learn Rust. Hence when you hire for a Rust job, you get a much better selection of candidates.

Yes. I would see readability and maintainability as the primary research focus of this PhD. Basic C-to-Rust translation has already been done and is not actually hard to do. It just does not get you much.

The Linux kernel is not allowing LLM contributions. It is allowing LLM-assisted (!) contributions and it has very strong quality control in places. Like "submit slop once, get ignored in the future" ones.

The licence angle is a real one, and the GCC folks make a very compelling argument for keeping LLMs out.

At this time, probably Gentoo. I have had a test-installation running for several years in case the stupid in Debian increases too much. But we will see.

Indeed. Could even go via USB-serial with CH340 or PL2393 and be dead easy and cheap. But a large part of the "business" view on technology is "do not pay an engineer if we can already make money off it, regardless of how broken it is". In a larger context, that mindset brings us things like climate change.

No, we do not know AT ALL how a human mind works.