61–70 of 208 posts
So Q* is just A* for neural networks?
Earlier quoted context omitted.
Doomers thought GPT-2 was too dangerous to release. I guess one can be Dalio successful by calling doom at every instant like Dalio.
Who thought that? I don't know of anyone who thought that and I'm in doomer circles.
Yann has been a refreshing source of reason and common sense with regards to AI safety, regulation and open-source. I wish we had more people like him and less AI doomer cultists.
People like his takes because he gives an authoritative gloss to what they already believe. But his points are usually lacking in argument or rigor. Anyone that essentially expects the public to trust them when it comes to the outcome of AI/AGI should be view with suspicion.
It is the same with AI/AGI. Anything closed source and having regulatory oversight is useless, decreases innovation, increases bureaucracy and will only serve those who wish to build a “moat” to further their hold on the technology.
I'm confused. I thought people were worried about the danger of some AI breakthrough? If researchers at OpenAI have developed an LLM more advanced than GPT4 which can also plan is that not potentially a worrying breakthrough?
It's just rumours. Everything I've read about that breakthrough sounds about as thoroughly backed in reality as a generic conspiracy theory. There may be a breakthrough, there may not, but nothing on the topic is convincing or worth reading.
I agree that the reporting on Q* has mostly been nonsense, and I suspect those articles were published just for the clicks, but at the same time this tweet makes zero sense because an AI model which can plan is somewhat concerning so perhaps those articles were on to something?
The leak about Q* feels like an olive branch to let the former OpenAI board and Ilya save a bit of face, probably part of the terms for Sam coming back plus it distracts from all the drama and puts a positive spin on things. >"The board didn't handle things well, but they were right to be concerned because OpenAI did have some sort of research breakthrough" Not coincidence that this leaks after Sam comes back, rather…
I don't buy this at all. Just going to leave a previous comment I wrote about people coming up with semi-outlandish reasonings behind why something occurs, when there is nothing wrong with the straightforward assumption (that is, the board was concerned about the direction of AI, which is wholly unsurprising given the board's makeup, the Q* stuff was just another concerning point, and there were obviously personality…
I love that reply: LeCun: “[Note: I've been advocating for deep learning architecture capable of planning since 2016].” Reply: “My understanding is Schmidhuber already solved that 10 years ago. Just no-one knows it yet.”
Earlier quoted context omitted.
People like his takes because he gives an authoritative gloss to what they already believe. But his points are usually lacking in argument or rigor. Anyone that essentially expects the public to trust them when it comes to the outcome of AI/AGI should be view with suspicion.
I mean I don’t think predicting the future is something that typically involves rigor. The outcome is pretty clear: whatever makes a ton of money. Probably a trusted friend in your pocket that sometimes helps you buy stuff. The most negative predictions are silly because they don’t involve making a ton of money for anybody.
Anyone have a good source for learning more about this topic?
I took an AI course in college back around 2015. Just a bit before AlphaGo. One project was to implement a simple Q-learning action/value system to play simple games, like Pacman. The crypto-bros-turned-AGI-experts on twitter are spouting the most uninformed, misguided garbage about this whole thing, it's quite amazing to watch. And I'm not saying that I am smart or an expert about Q* because I took an introductory c…
> The crypto-bros-turned-AGI-experts on twitter are spouting the most uninformed, misguided garbage about this whole thing, it's quite amazing to watch. After the LK-99 debacle where Twitter "confirmed" its superconductivity and breathless sheisters heralded the dawn of a new age, my new policy is "inverse Twitter": if Twitter thinks something is a big deal, then it's more than than 50% likely that it's unsubstantiat…
No clue why people try to come up with these obviously silly heuristics.