_cnt0

The actual real deal big boy models are so far beyond what your avg even extremely technical user understands. Unless you are actively watching following and using the models you just flat out have no clue just how fast this shit is sprinting.

You evidently do not understand the technology. LLMs inherently cannot reach AGI or anything resembling a thought process. And they scale poorly. Tons of compute are thrown at it to make the illusion marginally better. None of the inherent problems have been solved.

You could just as well burn money in front of a photorealistic painting in the hope it turns alive. No matter the amount of money you burn, it never will. Same with LLMs.

Thank you, that is reassuring.

I tried to make a joke. Looks like I failed.

needle-like fingers

Right ... you're operating your touchscreen with your dick again, aren't you?

Seriously, I wanted to open a PR to add myself to the list, but the fucker took the repo offline because he couldn't handle the backlash. Wussy.

At a previous employer they had actually good intentions with offshoring. Senior developers tended to be bound with maintenace of projects they had worked on over the years and the ranks couldn't be filled up fast enough to take over. Management wanted the most experienced developers to be availlable for new projects, so they decided to offshore some of the legacy maintenance to India. Didn't work out, though. The Indian company would throw ten people at things one of us would've done and they still managed to mess things up in major ways. My favorite memory is when they replaced a for loop with an if statement. Now, one could argue, that an if statement without an else branch is a special case of a for loop with one iteration ... but, this was not that 😅 Seniors ended up spending even more time on legacy projects to fix the crap the Indians did.

I'm in the middle of that bracket and am extremely negative about "AI" (LLMs, LDMs). I, and people born before me, grew up with technology. I sat on an Atari 1040 ST when I was 3 or 4. There's some stuff in the field of AI that's really exciting, like, for example, neural networks trained for pattern recognition to identify cancerous growths in early stages with much higher reliability than humans. LLMs and LDMs are not that kind of useful technology. I know how LLMs work and so I know that the intersection between their advertised and actual capabillity is tiny. There's no I in AI when it comes to LLMs. Yet they're the biggest investment bubble of all time. That bubble is going to pop when the average investor comes to the inevitable conclusion, that the technology cannot deliver on its promises. Until that economic catastrophy happens, it's fucking us in other ways on the way: wasting resources, negative impact on climate change, mental attrition (in those who rely on "AI"), depletion of seniority in all kinds of fields (using LLMs instead of training juniors), contributes to shifting of money to the capitalists, ... What's not to hate about "AI"? If anything my "tech savviness" makes me hate it more than the "unsavvy".