The real problem with AI

Ai in its current state is a bane of human's civilization. The technical details is very sound, but there are so many problems with it, which I dedicate this essay to explaining

The legal problem

A lot of data in AI's dataset are being used without permission on the author(s) and little is being done about it. Here's is an excerpt from https://githubcopilotlitigation.com:

By training their AI systems on public GitHub repositories (though based on their public statements, possibly much more) we contend that the defendants have violated the legal rights of a vast number of creators who posted code or other work under certain open-source licenses on GitHub. Which licenses? A set of 11 popular open-source licenses that all require attribution of the author’s name and copy­right, including the MIT license, the GPL, and the Apache license. (These are enumerated in the appendix to the complaint.)

In addition to violating the attribution requirements of these licenses, we contend that the defendants have violated:

GitHub’s own terms of service and privacy policies;

DMCA § 1202, which forbids the removal of copyright-management information;

the California Consumer Privacy Act;

and other laws giving rise to related legal claims.

And it's not just code either. A lot of media out there, be it art, video, music, etc... require at least attribute of some kind (if anything, it's common sense to give credits even if the author(s) doesn't ask for it). Yet, a lot of the times, no attribution is given at all: Bing's copilot does it, but many other like chatGPT or Poe doesn't give any credit whatsoever. This means that, at best, AI is living in a grey area legally and at worst, AI companies are just squeezing as much money out of this market as possible before they get sued into oblivion.

The resource problem

It would occur that this is a problem that is largely ignored, so here's a reminder that AI training require a lot of resource, particularly GPU power. It's as if cryptocurrency isn't a big enough problem already¹ that we now have to add AI to the mix to make thing even worse. At this rate, we will probably perish due to overheat and the ozone layer breaking before we get an AI that's capable of 1/10th of what human can do.

overpromises and underdelivers

I actually used to use AI - not for anything big, just to help me look up some math formula and solve my homework from time to time. One day, when I asked it to solve an assignment from a picture I sent it, I noticed one thing: It solves a different problem from the one I asked it to. Ultimately it was a math problem of similar type and thus have similar solutions, but in the end, it still wasn't the answer I was looking for. That's when I realized something: AI is faster than human, but a lot of the time, it is still wrong.

Ai bro say that ai will replace all sort of occupations, but frankly, all they have to show right now is a incoherent mess that serve no purpose other than to own the "ai denier". AI is cool in theory, but crumbles apart when tasked with actual, real-world problems.

Now, to clarify, I don't think AI is useless. AI has already allow people to be a lot more productive, and there a lot of things that AI could reallly help with (e.g screen readers). It's just that so many problems with AI that are either ignored or misrepresented, and I feel like some corrections and reframing of these problems could really help.