It's because they are doing more than simply assembling template code. As you say, that would indeed be achievable with modules and functions. But AI generated code is doing something a lot more complex.

There are people who dismiss AI as a "glorified auto-complete". To me, that just betrays a lack of knowledge about the AI of the past 5 years. It is like saying "The Large Hadron Collider is just a glorified slingshot", or "humans are just glorified protein factories". It is an application of neural networks, which has been around for many decades, but with the vastly improved hardware and improved algorithms, this has can now be done at immense scale. And the results are surprising - There is unexpected "emergent" capability, that researchers didn't expect.

Our instinct is to be protective of our own kind, and say that they are just machines, and they can never be creative like humans, or never have true feelings, or show love. And this may be true, but it challenges our notion of the essential difference between humans and machines. Are humans just complex machines, or do we have something magical (like a soul) which could never be replicated. The latest generation of AI feels like science fiction to anyone who was involved in AI in the 1980s.

It might be time for an update of the Robots Exclusion Protocol (robots.txt).

It currently lets automated crawlers know what they can access in the site. But it might be appropriate to have a finer tuned protocol which makes this determination based on purpose. e.g. I might be happy with a bot accessing my site for the purpose of search engine indexing, but not for the purpose of AI training... Or vice versa.

It would still be essentially a request rather than a physical restriction, but that's not nothing. A company can be asked whether they abide by this protocol or not. And large companies generally want to be seen as good "citizens".

Our history is littered with instances of technology ruining the previous world. The technology may actually be a net positive, but we can still acknowledge the negative side-effects.

We used to cook food from basic ingredients. It would take time and knowledge, but tasted wonderful. We now more commonly microwave frozen meals, which is quick, and requires next to no knowledge. But it has arguably ruined the culinary world. And yes, you can still find properly prepared meals, in the same way as you can still find properly written literature.

I think you make a good, albeit uncomfortable point.

If we accept that point - that a military is a necessity, and that it must not simply exist, but be effective - then we have to support the building and upkeep of such a military. But I still think there is a bit of leeway for the attitude in which we do that.

We can have an Eisenhower-like attitude, which I think achieves what you say, but in a sort of measured manner, alway cautious. Or we can do it in a bellicose, gung-ho sort of manner. Both might achieve an effective military, but somehow the former seems more appropriate, and dare I say, moral.

And the former approach often seems to be taken by those who have actually had experience in wars.

There are options which reduce the risk of this whole category of attack.
Two examples are (a) automatic zeroing of freed memory, and (b) inspecting and filtering suspicious HTTP requests.
These are not foolproof, but they reduce the risk of being attacked by an exploit of the next vulnerability.

The major reason they are not used is that they tend to come with a performance hit. But you'd think they might be useful as a temporary "switch on" option, when the alternative the company is considering is taking down the website.

There has been an increase in reports of cyber attacks resulting in the leaking of customers' personal data here in Japan, just over the last month or so.

The use of AI to both identify vulnerabilities and exploit them is presumably a period we need to go through, albeit painful. We come out the other side with more robust software.

It's a bit like if all the world were suddenly compelled to go for a health check... with the addition of bad guys who go around infecting people that are now known to have weak immune systems...

Perhaps, working for the creation of an American world-leading Super Intelligence, he can arrange a presidential pardon.
The law, after all, now seems to be a tool for those who can afford it, to wield against those who can't.

Wine has been going for a good 30 years. I have occasionally wanted to use it... although "want" gives the wrong impression. I might have had some hardware device that could be used with Linux, but the official Windows application had some particulr function that I wanted to use occasionally.

The experience was always fraught with problems, brittle, and wouldn't work again after having been left fallow for a few months. This was true whenever I tried it though the project history.

But you have to respect the project. Longevity!

Yes - I think it can be a good thing, and can give jobs to people who are unemployed. I don't think it is intrinsically a bad thing.

But - It is not as clear cut as "[...] products for American consumers here on American soil.". The idea is what is sometimes referred to as "near-shoring" - a sort of compromise between off-shoring and on-shoring. It does have the benefit of new jobs in those "near-shore" poor states, but it tends to keep the glamorous jobs firmly situated in the wealthy states, and there by reinforcing the reputation of both states.

And if you are moving from off-shoring to near-shoring, then of course you have to go and tell all those off-shore workers that they no longer have a job. If the firing manager is a keen America Firster, then he can probably console himself pretty quickly about that, but a "people are people" person might need a little longer.

Immorality and corruption are nothing to be ashamed of in Trump's mind. Being concerned about that sort of thing is for "losers". On the contrary, they can be carried out in the open and bragged about.

This is why, having been passed over for the Nobel Peace prize, Trump called the decision an "indelible stain on the country of Norway", not because he is unaware that the decision is not made by the "country of Norway", it's just that he finds it difficult to comprehend a government that won't interfere in the organisations within its borders.

It just baffles me that a country so proud of its founding concepts - separation of powers, no one is above the law, aversion to tyrants, etc. - could select this person, twice, knowing his character. It's utterly baffling.

Is this a case of on-shoring assembly, or off-shoring Mississippi...?

Hundreds of lawyers rubbing their hands thinking of ways to put an asterisk against "own".

That's an interesting point. I can see how companies might register this as a risk, and avoid it on that basis, even if the legal conclusions have not yet arrived.

To clarify, when you describe LLM output as copyrightable, are you implying the copyright holder is the LLM provider, or the owners of the original training data?

In discussions of copyright on this topic, the theme is usually a complaint about LLMs being trained on data created by humans, without the permission of those humans. They might have a case here, although (a) it would be a difficult one to unravel, and (b) since the LLMs generally do not reproduce raw training data verbatim, they arguably only violate copyright as much as humans do when they read a lot, digest that information, and then compose something "new".

But if - as you may be suggesting - the AI companies might decide that they are copyright owners of LLM output, then we are presented with a different kettle of fish. I wonder if any of the small print that comes with those models implies that... It sounds difficult to enforce - being so widespread, especially retrospectively, and difficult to prove. But I agree, a risk is a risk, and companies often avoid that sort of risk regardless of the rights and wrongs.

Caution is required, as it is with any submission to the Linux kernel, but I think it's sensible for Linus to be open to the idea (albeit with his guns-ablazing turn of phrase).

At the other extreme, we need to be careful of the knee-jerk reaction of writing off all AI generated code as "slop". Sometimes this can be no more than a sort of "signalling". If I label code as "slop" I am really just letting you know that I am a programmer at a level that can discern such things... even if I am not. (Since feeling I need to let you know betrays an insecurity).

In my limited experience, AI can generate good code, particularly for small, well defined cases. And even when the first draft is not good enough, you can refine the requirements and get it to improve the code. Just as you might with a human.

That's a good point.

A common criticism to any sort of benefit scheme it that the recipients are not deserving. But of course we often decide to expend additional resources on people who might not be deserving - If our child goes off the rails, we will do what we can to get their life back on track. We can justify this, saying the child is one of our own, while the homeless stranger living a mile away under a bridge is not. This defense may be common sense to most, but in using it, we start to sound a bit like the scribes and pharisees...