Jump in the discussion.

No email address required.

chatGPT is as far from AGI as the first wheels were from airplanes

Jump in the discussion.

No email address required.

I wish journoids would stop calling LLMs AIs

Jump in the discussion.

No email address required.

I think they are more than LLMs. I actually think ChatGPT will use an LLM to write Python scripts that model the questions you ask it. I've been very impressed with how it handles weird abstract math problems, which come up in my life surprisingly frequently.

Jump in the discussion.

No email address required.

It's trained on (unpublished) textbooks and academic papers, it's not shocking that it can autocomplete an answer that sounds plausible to your abstract math questions.

Jump in the discussion.

No email address required.

I mean it will come up with a simulation and tell you the answer if it's a problem that's not tractable to actually solve. I asked it something that I'm pretty sure would have taken longer than the life of the universe to solve and it told me to just simulate it and stop pretending to be a mathematician (which was the correct answer).

I'm trying to think of an example that won't out me as a dramatard to my coworkers if they also lurk here.

Jump in the discussion.

No email address required.

At their core LLMs and ChatGPT use the exact same framework, there is no intelligence there, it's just mapping the results to a curve

Jump in the discussion.

No email address required.

but calling everything AI is how you ELEVATE SCALABLE ROBUST SOLUTIONS

Jump in the discussion.

No email address required.

They're entirely different concepts too. LLM's are basically just probability based random number generator's, it's the equivalent of a bogo sort. To get 100x what we currently have we'd need a million times the computing power

Jump in the discussion.

No email address required.

youre gay

Jump in the discussion.

No email address required.

Yeah they aren't advanced learning networks that can self evolve it's an adaptive multi billion vector answering bot. All it does is basically what ai art does for words. It doesn't know what an apple is, it just makes something round and red when you write apple.

Jump in the discussion.

No email address required.

what's your brain then?

Jump in the discussion.

No email address required.

Wrong. AGI by 2029. Some people are saying AI is scaling up at a rate of moore's law squared right now. Moore's law states doubling of transistor counts every two years. That means AI is becoming 8x better every two years or 4x better every year. Which means in 5 years it will be 1024 times better than it is now.

Currently AI is on the level of a mid tier graduate.

Double that for a good quality graduate. Double that for a college Professor. Double that for an actual job. Double that for low level researcher. Double that for mid tier researcher. Double that for high level researcher. Double that for CIA operations.

Which is 128 times more powerful AI.

AGI by 2029 guaranteed.

Jump in the discussion.

No email address required.

>mid tier graduate

I don't think mid tier graduates nowadays are even as good. At least I'm not being r-slurred

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.