No, I think the likes of Musk/Bezos/Zuckerberg will prefer to live where ever they like, protected by an army of Robots which, unlike the security personnel they currently need to employ, will not object to whatever atrocity they are asked to perform.

I'm honestly trying to understand how UBI would be supposed to make a difference: Let's assume the extreme case where Robots produce everything vastly better and cheaper than humans ever could. Let's assume the government is willing to pay out arbitrarily high UBIs. Why should money in the hands of those not owning the Robots motivate those owning the Robots to give them anything? Those owning the Robots (including the Robot army to defend themselves) already have everything they want, and money from some no longer needed humans is not really of value to them anymore.

So instead of UBI, a government interested in the welfare of its human population would rather have to somehow provide the amount of Robots, land, and natural resources needed to become self-sufficient to each of its citizens. Which will, of course, meet fierce resistance by the original Robot owners, who certainly have no interest in sharing resources with just everyone, and will let their Robot army fight for them.

I am sure there a plenty of tests. Written by exactly the same LLM that also wrote the code. And therefore expects the same defect output in its tests that its code generates. It was never a good idea to have the same programmer write both the code and its test, whether human or LLM, but now there are far fewer distinct LLMs than there were distinct humans to do the job in the past.

Yes, damning food just because it is "processed" is stupid, as there are many forms of processing that improve food health-wise, and yes, there is a lot of hysteria around "what diet one eats" with little evidence backing it.
What they should really say could use an abbreviation: "Food processed by greedy corporations that improve their profits at the expense of making it more unhealthy for the consumer should be avoided to become the main component of one's daily intake."

The small overall score difference between Claude Opus 4.7 and DeepSeek 4 in contests that are based on mostly English speaking participants scoring the results could well be the other way round when doing such contests among a mostly Chinese speaking population.

That is why I do not connect the Xbox to a network. And will certainly never buy any future one that refuses to work without network connection.

Yes, for the price of a new gaming PC they will send you one, outfitted with an "Xbox"-sticker at the front. Then you will only need to buy once again all the games you have on your Xbox discs, and you are good to go again. Ok, some of those Xbox games you previously bought won't be available, but that is just to enhance your experience.

I think nobody at the top of Microsoft really expects or even wants to re-enter the console market. They just want yet another marketing bullshitter join the attempt to mislead people into believing that that Windows pig painted with some Xbox-branding lipstick is a gaming console.

... to allow for the use of LLM-based bots to read texts or to do math, and just like that scores will dramatically improve. I'm sure there are certain companies willing to sponsor such a change of scoring rules / education goals. Remember, as one famous Oligarch stated: “I don’t want a nation of thinkers, I want a nation of workers.”

... not for Microslop, who are just doing their business as usual, and won't squeeze any less money out of their vendor-locked-in victims.

And even when they notice that vibe coding does not work that great, they will still try to move expenses away from wages towards tokens paid to some LLM hoster. And once they find out how expensive that gets over time... well, they probably have been replaced by LLMs themselves at that time.

This is indeed an embarrassment. Just like some newspaper recently reprinted Espressive touting about "AI on ESP32" micro-controllers. Sure, you can make expensive API calls from those, too.

Your "no true Scotsman" argument isn't compelling. If people used LLMs only to "lookup facts" or to "translate one language into another", there would be way less reason for concern (well, they would still need to retain the ability to check the results, but that is another topic). But people use LLMs for everything, en masse, from "applying math & developing algorithms" to asking what time it is and what to eat today.

The problem with LLMs is not that people are losing the ability to do one or another specific thing, the problem is that they are unlearning to think as a whole. If you practiced juggling or playing piano at some point in your life, and later, due to a lack of opportunity to practice, unlearned it, it may be a pity, but no big deal, and you can still move your body to do other things. But if you stop moving your limbs entirely, even if only for months, the muscles atrophy, and you will have a hard time doing anything requiring body movement from that moment onward.

The scary thing I observe in many colleagues, right now, is that they really atrophy their brains. It does not matter that they forget about certain API calls or programming language features, those could be picked up again quickly from some documentation again. What matters is that they stopped having any sophisticated thoughts, they outsource all thinking to LLMs, and start using them for more and more mundane tasks while becoming more and more uncomfortable when asked to think on their own. They cannot answer even simple questions about "their" work results anymore, because the results aren't really "theirs", but fell out of some LLM.

I hope some people will be able to use LLM responsibly just as a tool, rather than as a brain substitute. But I am concerned many people will atrophy their brains for good.

They'll get their revenge at the day of the Triffids.