Future Timeline

It's Demis Hassabis suddenly saying "ten years" after spending so long saying "several decades" that gives me pause.

TrueAnimationFan wrote: ↑Wed Apr 13, 2022 10:04 pm "The artificial intelligence system disappears...in a quest to search for more."

What's that supposed to mean? That it launched itself into space to look for other alien civilizations that collectively uploaded themselves into their AI systems? That it concluded existence is meaningless and then shut itself off?

I always interpreted it as the AI desiring to break free of Earthly limitations and explore the cosmos to maximize understanding. Of course, it's really just a speculative analog of Creation.

The Average Joe, the kind of person that thinks something like the Oculus Rift CV1 is still 100 years away, is going to get reamed.

Yuli Ban wrote: ↑Thu Apr 14, 2022 3:38 am

The Average Joe, the kind of person that thinks something like the Oculus Rift CV1 is still 100 years away, is going to get reamed.

As is tradition. Hell the Average Joe refuses to listen about developments in things like the digital dollar and more that are happening right now if you told them about it. As unfortunate as it is it is a very hard truth I had to accept that a lot of these folks are just dumb. All the internet and tech did was expose how bad the situation is with dumb people.

The way I see it, the first proto-AGI will arrive as soon as an AI lab manages cross PaLM, DALL-E 2, Jukebox, and MuZero. The strength of the model doesn't matter since we're talking about generality. It could be GPT-2, CLIP, MuseNet, and AlphaGo in terms of strength, though stronger is better.

This can't be more than a few years away. All the pieces are already there.

The Average Joe can be forgiven for not being prepared. I don't think anyone is.

More and more I think we are living in a very specific time period where we can both actually observe exponential growth in AI e.g. GPT3 vs GPT4 (when it is released) but secondly at the same time AI feels really slow.

It is realistic to say there is a good chance GPT 4 will be twice as good as GPT3 so basically one doubling in two years akin to moores law. Yet at the same time AI progress feels really slow whist we live it, waiting 2 + years is a drag and like GPT3, GPT4 will still be miles away from human level AGI.

A cool interview about the rise of AGI.