Ilya Sutskever | AGI will emerge next year and will revolutionize the world

[KGN_VC]

Thanks for the videos. But here is some constructive feedback; you really really need to stop adding disingenuous titles. “AGI will emerge next year and will revolutionise the” it’s just straight up click bait and bordering on desperate. It’s really not a good look.

[HouseOfSynister]

According to ChatGPT: Research in Mathematical Sciences on December 14, 2017

The talk by Ilya Sutskever on meta-learning and self-play took place at the Fields Institute for Research in Mathematical Sciences on December 14, 2017. This event was hosted as part of the 2017-2018 Machine Learning Advances and Applications Seminar series. During his presentation, Sutskever discussed meta-learning, which focuses on training systems that can quickly learn to solve a variety of tasks, and self-play, which provides an incentive for indefinite improvement by having algorithms compete against themselves

[johnaldchaffinch]

This has to be old, I can only judge it based on the fact that he has more hair on the top of his head than the last few years. I'm eager for a new video from Ilya, he'll have more to say than anyone...

[cesarmateogonzalezrodrigue]

I have watched a lot of you videos, but several times I have the same question, when the interview or conference happened? I like to be up to date. Are they published once they happened?

[drawnhere]

Maybe you could time stamp at the point where he mentions at AGI will be here so soon. The talk is extremely dry and too far into the weeds. Thanks...

[sophontec]

You are correct, this is the good title for the talk, although it is old. I guess it is exactly the theory behind the Strawberry or whatever naive AGI algorithm. Just one thing missed, the representation of logic and reasoning. Maybe another neural network works.

[simontv]

Did I violate a YouTube comment rule? Simply put, thank you for the video and peek inside the future

[mrd]

IIYA says unlimited amount of meat available at chioptle next year😅
And thats not clickbait either..

[AGIBreakout]

Clickbait.. don't watch... older video.

[galaxiesmerge_original]

As I watch this video it is clear how old and unoriginal any of the ideas are. Of course I worked for U of Toronto also and enjoyed that. I don’t find it interesting so far but I see him sucking up the buzz words like “meta-learning” - what will the have next? Meta-Poesis? Meta-Semiosis? Meta-Episteme? The Greeks gave the buzz-word kids a very useful tool to sow the seeds of ignorance at the fastest rate anywhere. How do you even begin to define what the ridiculous phrase “superhuman learning” means? What in the name of Sciience which Ilya purports to practice, is the meaning of “superhuman”. I’m dropping off. So sad. I thought he had something. But he has nothing. Good luck to his efforts. I hope to revisit in a year or two and see what his actual contributions are.

[galaxiesmerge_original]

It is old. It is real old. There is no connection to entanglement, no understanding of the relationship between classic constraints and non-classical structures leading to quantum-like behaviors. So this is very old. Very old and not very new. It’s sort of a tourist survey. But so I think a nice but watered down, weak and old rendition of the truth of the paths of pursuit to generalized learning. In fact, the correct problem is to be stated as: “The Problem of Learning How To Learn”.

[michealhaines]

Old video, bad title, wont be watching your videos again