True. Maybe that sensation of "wow this just saved me two weeks" happens often enough to leave an impression, but only in narrow enough areas where it doesn't make much of a difference in the grand scheme of things.

Heh. Yeah, I would just ignore it, it's far beyond a good faith discussion. What you're saying should make obvious sense to anyone.

That's also true but a different statement :)

I guess you don't mean AI in general, but the current LLM-based wave of AI?

en.wikipedia.org/wiki/F%C3%BC...

There's no room for laws or courts in the Führerprinzip

Oh yeah 100%. If it's anything more than a short-lived or uncritical tool, my rule is that I absolutely have to understand every generated line, and I go much more for in-line tabbing etc.

For quick works-for-me stuff, I'm happy to vibe code larger things.

You're right we can't, but on a subjective level it sometimes gets hard to discount if you have enough moment where you go, "Wow, this just saved me 2 weeks of work".

There's a lot of opposite moments as well, but you can get better at avoiding them by getting a feel for how/when to use LLMs.

Yeah, I think the most important thing for junior developers to understand is that they shouldn't just trust LLM output and always still try to understand what's going on. But it's great for getting pointers in the right directions, discussing different architecture/implementation options, etc.

Yes, I read that and know that, but that's very different from hacking the tool itself. If I get a Windows user to click on and execute my trojan, I haven't found a vulnerability in Windows, but in the user. That's quite tool-independent. VS Code asks about trusting code all the time as well.

You can still call it hacking if you like, but then the title is still at least misleading, since the tool wasn't hacked, but a person was. Still good to be more resilient against that, but I dislike misleading titles.

"To execute the hack, he only had to convince an Antigravity user to run his code once after clicking a button saying his rogue code was “trusted” [...]"

Ok so not actually a hack at all, but social engineering?

Ugh. Ok, reading www.theguardian.com/technology/2..., it says:

"MEPs passed a resolution on age restrictions on Wednesday by a large majority. Although not legally binding, [...]"

It is encouraging indeed. I've just seen him survive "impossible" things too many times, but let's hope 🤞🤞🤞

🤞🤞🤞

The problem is that people have been saying that he's deeply weak and at the end of the road for around a decade now, and somehow he always manages to weasel himself out of any situation, again and again. Of course, I'm hoping that this changes at some point, but still...

Yep, agree on both!