I’ve been on Lemmy for about 2 weeks now, and I’ve noticed a trend:
The VAST majority of posts that mention AI in any manner are some dig or criticism or some other negative commentary, and the rare ones that have anything positive to say about it almost always have negative whatever-Lemmy’s-version-of-karma-is.
I get that AI isn’t without its problems, especially Grok with that “Mechahitler” nonsense a bit ago, but there seems to be particular vitriol here. I’m genuinely curious to know why people hate it so much here.
Because it’s trendy and it’s a new thing people can feel outraged about together. And when you really drill down into the arguments, the problem is invariably with the way technology is used under capitalism rather than technology itself.
I somewhat agree with you, the tech is amazing. But I also think that the current shape of “AI” is what it is due to a massive unsustainable investment directed in the vast majority by ill intent. The technology would not be what it is today without it being used for what is hated for. If you “fix” the hated parts then it becomes something entirely different to what it is today, if anything at all.
I think that the perspective that the tech is is hated because its trendy looses weight very rapidly when you realize that it would probably just not exist if not for the insane power-hungry assholes that are shoving it down our throats every damn day.
I don’t really agree. I use this tech for coding, doing translations, speech to text transcriptions, extracting data from PDF documents, and none of these use cases would be different if the tech wasn’t commercially driven. I also disagree that it wouldn’t exist if it wasn’t developed under our current dominant economic system. AI research has been around for a long time, and has been done extensively in socialist countries like USSR.
Its not that it wouldn’t exist, its that the current form is shaped by the profit incentive. The fact that is a useful technology doesn’t mean that the AI we are being forced to use is like that due to a capitalist mode of production. Coding help may be useful to a point, but its only so thanks to a massive level of compute power that wouldn’t be available under a reasonable mode of production.
I wholeheartedly diasagree with the idea that a socialst world would have gone with this as a priority to the point of making it something like what we see know. Would it be developing it? For sure! Would it have anything related to its current most extended use cases? No way.
The whole trope that LLMs need absurd levels of energy use has not been true for a while now. People latched on to this idea because early models were hideously inefficient, as is the case with pretty much any new technology. Today, you can run local coding models on your laptop that surpass the capabilities of frontier models needing whole data centres to run just a year ago. You no longer need an inordinate amount of computing power to run any of this stuff, and performance gains haven’t stopped. There’s no indication that we’re close to any sort of a limit here.
Also, nowhere did I say that a socialist world would have developed it in the same fashion. I’m merely pointing out that it would have been developed, and there would have been many existing use cases which I listed which have little to do with commercial incentives. I get the impression that you’re conflating hype with the actual legitimate use of which there are plenty already.
Finally, there is really nothing stopping people from developing this technology in open source fashion. And that’s the way to decouple this tech from commercial incentives going forward. There are already open models to build on, and that should be leveraged to develop completely open alternatives which are community driven.
Well hardly anyone here would claim that LLMs and machine learning are bad tech.
I think you would be surprised. Read on and be astounded.
Nailed it in one! 👏