Because for the past 30 or so years, it has worked very well for MS to keep their main products barely useable, rely on lock-in and chase the next big thing so they can get their dirty hands on it early and lock more people into more products.

That is exactly wrong. Beta versions are there to find bugs. If nobody complained about them, they won't be found. So if you try beta versions, please DO complain about bugs.

I guess "vibe-something" is going to be the anti-word of 2026. People are slowly waking up to what it actually means to let the AI do the work.

I'm not dissing AI, I'm using it extensively myself and there's a few AI whitepapers with my name on them. But like any tool, it can be great when used correctly and ruin your day when not.

That doesn't mean what they do is automatically legal.

I'm pretty sure there's a demand for "murder my husband", or for "can someone please drown my neighbour". And yet it's illegal to offer such services.

Also: Voters. Smokers are still a fairly substantial fraction of the population, enough to swing a vote, especially if, and that appears to be the trend in most western democracies these days, there are two opposing political sides roughly evenly matched.

I mean, does it not strike anyone as a very weird coincidence that we have almost perfect 50/50 splits in so many countries?

No, but they do everything BUT force to make it the most attractive option. Just as one silly example: With wages and prices as they are, having both partners work full-time is basically required unless you're in the top few percent of earners or inherited wealth. So who's going to cook? After a long work day? Convenience food is the obvious choice. You are not being forced, but unless food is a high-priority item in your life, you are very much steered into that direction.

The point was not about them being the same technology. The point was about them both being technologies that do a better job than a human in general, but may do a catastrophically worse job than a human in specific situations.

Also, you don't seem to understand that current gen AI is algorithmic. The "self-learning neural networks" are algorithms that will produce the same output from the same inputs. Either you have not actually studied computer science, or you did not understand some of it.

While there are circumstances where you can use them without actually quoting something, there are certainly situations where the expectation is that you are actually quoting something. For example when you are claiming some specific regulation or ethical rule and then you put something in quotes appearing to be a quote of the rule or, for another example, the sentence I have copied above where you appear to be quoting me, but just made the sentence up. In any case, whether or not it's a quote, it's still generally expected that you'll give a citation.

Sorry, but you're basically babbling like an idiot and you're still entirely missing the point for some weird tangent that has nothing to do with what I was saying.

I mean yeah, it's unfair to blame him for more than about a quarter of it.

You are completely failing to understand my argument. I am completely in favor of robotic surgery and I think that robotic systems are capable of being far better than human surgeons. I am talking about autonomous robotic systems, however. The robotic surgery systems you are talking about are not autonomous, they are still very much guided by a human surgeon. My actual argument is about what you do when safety critical autonomous systems do things that are unpredictable. For another example, consider semi-autonomous flight systems like the MCAS system in the Boeing 737 MAX. Technically the MCAS system handled the unstable flight dynamics better than a human pilot. So it was a semi-autonomous flight system that made the plane safer than if it were flown directly by human pilots without that system. However it also force-nosedived two different flights into the ground killing hundreds of people. The argument is simply that, even if an autonomous system makes something safer statistically, it does not excuse the system having avoidable safety lapses. You still identify those lapses and fix them.

Aside from your reading comprehension fail over what I actually wrote, I would love for you to actually cite the quote you put in your post. If you were putting in quotes like that, I assume you were cutting and pasting directly from a source, so it should not be too hard for you to cite the source.

There is no real singular definition of ultraprocessed foods of course, which is one of the fundamental problems in discussing them or in passing laws and regulations to control them. Also "emulsifiers and stabilizers", so... mayonnaise?

Right, which makes nixtamalization ultraprocessed rather than just processed. Maybe you keep quicklime in your kitchen though. It's not typical for most people though.

In agreement with most of that. Just need to point out that, with good situational awareness, you can brake and continuously track the vehicles around you while you do it and, in the worst case scenario, stop braking and accelerate again if necessary. If I can do it, the AI should be able to do it. Once again, we don't know what the AI knew or was "thinking" or attempted or did not attempt in this case. I would like to know.

You can point that out, and I'll point out that the original argument from registrations_suck is that only an agency at the federal level should be able to sue in such cases. That was one of the reasons I specifically chose an example from China, which has the kind of supreme federal power that registrations_suck was suggesting should be the case here. I mean, another reason was because it was such an egregious and horrible example. However, all the reasons you listed where the federal government might cover things up to avoid embarrassment (and protect corporate leaders) and not do anything until it becomes impossible to cover up are exactly the reasons why I don't think a federal agency directly under the control of the executive branch of government should not be the only entity with standing in such cases. Thank you for providing that perfect argument.

Okay... thanks for the factoid.

Exactly. Also, I am left to wonder whether these sorts of things that can actually be handled by incremental improvements on the self-driving models or not. The field of AI in general is plagued by leaps forward that seem like breakthroughs but come with fundamental limitations that can't be overcome without basically starting from scratch.