I agree. And bad or irresponsible coders are bad or irresponsible coders and their code should not make it into anything critical.

That does sound good, but it amounts to solving the halting-problem and that one is unsolvable. I do not think this task is something that AI can sufficiently approximate to be useful. You would never know how much it missed and it may also hallucinate differences that are not there or are there but behave differently.

I disagree. Sure, they may remain as toys, but to kill their viability for real work needs very little and there are a lot of known problems already.

It depends. When you take it seriously, enjoy it and have talent, an applied CS or IT degree makes you an expert of which there never will be enough. If you "just want a job", CS/IT is a really bad choice though.

Management is still fighting the idea that engineering works needs qualified engineers (and a good applied CS degree is effectively an engineering degree even if typically not called that), but my take is that when the current AI insanity collapses, it will be another step to the general recognition that IT is really hard when you need to do it right. And at least Europe is on the way to warranties (software warranty for end-users is already there) and liability (NIS2/KRITIS is a preliminary step) for IT products and services. And that will mean things get a lot more professional than the currently so common half-assing of things.

Hence I would anybody with the interest and talent still encourage to go into it.

It needs both: An engineering track that is proper engineering, and teaches all the advanced and hard to do stuff. Say 10% of the graduates. And a "technician" track that is more like a trade-school and teaches how to do all the well understood standard stuff right, like system administration, business coding, network administration, etc. Say 90% of the graduates.

Incidentally, for all engineering except IT, Europe has that. And ways to go from "technician" to "engineer" if you find you want that and have the talent.

The Great Barrier Reef is toast. This is just an effort to make that less obvious.

That is a possibility. It would have been relatively easy to find out he was after underage girls and then use that to recruit him.

Trump is definitely trying his best to keep the files under wraps and he did not do that before. I think his promise to publish them was genuine. Now his fear of them getting published looks genuine as well. Which probably means that Trump did not rape any teens, at least not with the help of Epstein. But something they found in these files in the last few months has Trump deeply scared.

Well, German red tape is bad enough to turn people away that Germany actually wants. Obviously things can get even worse, no argument.

Interesting. I did expect pretty much the described signs of rot, but not this pronounced in a mainstream product yet. That the others still work is probably just a sign that they are more conservative on the training, i.e. more outdated on facts.

Well, if these tools are hit and miss like that, they have already become unusable because they can degrade at any time and you cannot depend on anything they produce. And no, they will not be getting better. They will only be getting worse now, except for specialist LLMs that have been trained with manually selected training data, which will be exceptionally expensive. It may still make sense for some applications where a lot of relatively uniform documents are processed, like in evaluating tax filings and the like.

Nope. Unless you plan to throw it away anyhow. Oh, sure, you may get something out this way. But it will (a) take more effort and (b) has a high risk of some hallucinations ending up in the final product.

That is the only point I sort of agree to. Within limits.

I do see that, say, prototyping the look & feel of a GUI is a task where an LLM can help. But for anything that must be done right, you are better off using that LLM as "better search" only and even that may slow you down or introduce errors.

They were thinking "Dollars! More Dollars! Many More Dollars! And who cares about the customer!" That is what Microsoft usually thinks. Lets hope this time it comes back to haunt them.

I expect real escorts (the hooker type) would have been more effective, because they typically have a good radar for when something feels off. They would have been about as effective on the tech side, because it is essentially impossible to monitor what something competent does on a computer keyboard. I know, I have tried.

The Chinese will have anything worth knowing from these MS customers now. And they will have placed countless backdoors. "Escorts" like the ones used by MS are completely worthless. I had the opportunity to observe that in person and to run some experiments a few years back (me and my boss tried to "supervise" each other, complete failure). The customer still decided to go with this flawed idea. And a few months later a developer placed a backdoor and they only noticed a few weeks later on a traffic analysis. And that developer just wanted easier access for his work, so no/low criminal energy involved.

I am sure MS knew that this scheme is just a lie and does not work. I am also sure they did not care. I would call that treason in the case at hand.

Oh, and the cleanup? They would have to throw everything away and rebuild these systems from scratch. That will not be happening as it would be a multi-year effort. and hence the Chinese will retain access to everything. I am sure the Chinese intelligence people that did this attack are still laughing at this incredible level of stupidity and sheer incapability.

I disagree. A degree in circus management may be of some use.