Indeed. The last few AI hypes were bad enough, but this time the greedy assholes behind it have lost all sense of proportion. Well, the crash will just be even more spectacular.

I fully agree. This AI hype is more bombastic than the last few ones and it looks like it will actually deliver even less.

So far, LLMs can only do "cheaper" and it always comes at a reduction in quality. There is no reason to expect this to change.

But seriosuly, the signs of the current AI hype comming to an end are getting stronger.

There is a bit more to a language that just "not making errors". I guess you never really learned a language.

Sure, the occasional dumb narcicist may quite like LLMs, but most people want a bit more meaning and genuine interaction in their lifes. Chatterbots cannot do that. Such a surprise.

Yep, same here. And the complete lack of vendor liability for software, no matter how incompetently done and no matter how much damage happened as a result of that incompetence. We can only hope that Delta wins decisively at the totally clueless fuckups at Crowdstrike and that this serves as a reference case afterwards.

Hahaha, thanks. AC = fuckup with no insight and no honor or decency. They have standpoints so broken that they are not worth even considering and they are too cowardly to even give you a pseudonym to respond. The value of people does not get much lower than that.

My students tried "ChatGPT, add to this small, simple, 30 lines game-of-life in Python. Complete failure. I expect "specialist" models may be able to do this miniscule, easy task or not. But the level of complexity at which LLMs do not get it anymore and consitently fail is very low.

Indeed. Not voting is also voting. Obviously, the US has made it hard for many people to vote and that is one factor in the current problems. But even if it is hard, you become complicit if you do not vote.

Having acquired some experience (unfortunately) with trying to teach some people from that group, I completely agree. Never again. These people are not into understanding things.

It is however, the reason why "Teach all kids to code" edicts aren't all that successful.

Fully agree to that. Obviously, the very idea that you can teach everybody to code is also hugely disrespectful and insulting to those that can learn to do it well. My personal theory is that the MBAs and other "business" idiots feel inferiour and so they claim that the skill is trivial.

That idea is too difficult for economics graduates and MBAs.

I agree, except that the "some will work well" part will likely be rather small and completely offset by the others.

Only that it will not. These large companies can do it because they have high momentum and even sacking all their coders, for example, would only kill them slowly. But they will find out that this does not work. And then they will be struggling for a long time to undo the damage. If they can.

The only thing that will happen for the ones that go this route is deep enshittification and then possibly corporate death. AI cannot even successfully replace a worker that is somewhat dim. Oh, and all the good engineers that got sacked will remember ans spread the word. Good luck ever hiring competent people again.

I mean, car makers generally understand safety engineering and hence that tactile feedback is critical to not distract the driver. Where these abysmally bad decisions made by non-engineer "managers"?