150
submitted 2 days ago by yogthos@lemmy.ml to c/socialism@lemmy.ml
you are viewing a single comment's thread
view the rest of the comments
[-] balsoft@lemmy.ml 19 points 2 days ago

There are currently other problems with AI but they are much easier to solve when those profiting from them don't own the government. See: China's regulation on AI chatbots

[-] wheezy@lemmy.ml 13 points 2 days ago* (last edited 2 days ago)

Since this comment is about the edge cases I'll add my own here as an example. It's not about "AI" in how most people think of it.

One example of "AI" that is more efficient and more accurate than traditional algorithms is in OCR (Optical character recognition). Traditional algorithms rely on templates and are very dependent on configuration. OCR is basically when you take a picture of some text on a sign or a page. The conversion of those pixels of text in the real world into text that you can copy and paste on your phone. Basic shit we've had long before "AI". Modern machine learning has improved this drastically. The same field that brought us "AI".

The "AI" in this case is really an evolution of machine learning in applications of computer vision . When we decided to call that "AI" was when the capitalist decided it was good marketing.

I worked on machine learning in college a decade ago. It is an area of computer science that has a lot of potential to make computer systems more efficient and more accurate when interacting with the real world and applied to computer vision models.

All of this shit has been thrown out of the window for chat bots that can be marketed and sold as a consumer commodity and experience.

I say all this, because "AI" is not really anything bad in its foundational principles in computer science. The insanity of where we are today is entirely because of the ownership of the means of production.

I tried to find the comparison of OCR algorithms I read in the past. The "application specific AI" models use less computational power and are more accurate than previous template models. This was a year ago I think. But, because of how awful the search is now (throw AI into something that doesn't improve from it) I can't find it.

DeepSeek-OCR and some other algorithms were compared.

Anyway, the point of this is that there are cool improvements in computer science that use machine learning. As someone in CS I fucking hate the bloat on this industry that has made a generation of society hate computers and tech. Capitalism is a fucking plague.

[-] yogthos@lemmy.ml 10 points 2 days ago

Yup, every technology can be abused and has negative effects. It is the job of social structures to make sure it is used in a way that's actually beneficial to the society at large.

[-] AlteredEgo@lemmy.ml 2 points 2 days ago* (last edited 2 days ago)

It's almost shocking how fast and smart China regulated these AI girlfriends (but at what cost lol). Companies can still do it, you just have to do it responsibly.

this post was submitted on 27 Jul 2026
150 points (97.5% liked)

Socialism

6903 readers
194 users here now

Rules TBD.

founded 7 years ago
MODERATORS