Rationalist bozo apologizes for previously being right.
On being wrong about AI
71–79 of 79 posts
Re: On being wrong about AI
#72Earlier quoted context omitted.
We have self-driving cars already, but they're not evenly distributed ... and they have an unsatisfying amount of human input behind the scenes.
The report that came out was about Cruise. Waymo didn't have the same "unsatisfying amount of human input behind the scenes".
As far as you know, yet.
Re: On being wrong about AI
#73> Eliezer himself didn’t believe that staggering advances in AI were going to happen the way they did, by pure scaling of neural networks. I'd say its more than just scaling, only the smallest forms run on a modern phone, slowly, with caveats. Its far beyond anything reasonable. I still find it weird that there are probably multiple datacenters worth of compute just generating text.
Text, waste heat, and CO2 emissions.
Re: On being wrong about AI
#74Just a small question about that: If we train a generative AI on AI text, is it becoming better, worst, or the same? Because if GenAI2, trained on GenAI1 production, is less useful or as useful as GenAI1, we can be pretty sure it isn't creating anything.
It will become worse, to a greater degree than you would naively expect (Model Autophagy Disorder).
Re: On being wrong about AI
#75This article is absolutely pointless. The statement was from 2009. Back then, neither side had any real evidence in favour of their argument. Deep learning and AI were still purely academic experiments, with zero practicality. And it had been like this for half a century. Back then, even the best models couldn't tell cats from dogs. Today every high schooler with a laptop can solve that problem thanks to abundant har…
The only worthwhile thing it mentions is Ray Kurzweil, then discards it out of hand without actually considering the methodology because it must be unsound to be so precise:
> The few who did predict what ended up happening, notably Ray Kurzweil, made lots of other confident predictions (e.g., the Singularity around 2045) that seemed so absurdly precise as to rule out the possibility that they were using any sound methodology.
The methodology (ignoring all other detailed knowledge) is roughly our progress with computers and software being exponential for its entire history. What lay people don't get about literal exponential growth[0] is how slow it is (vs linear, polynomial) before it 'hockey sticks'. I assume everyone here knows that and the Wikipedia link is only for the pictured graph as reference. If each unit on the x-axis is a decade of computer hardware and software advancements, we may very well be only two ticks away from the singularity.
Re: On being wrong about AI
#76This article is absolutely pointless. The statement was from 2009. Back then, neither side had any real evidence in favour of their argument. Deep learning and AI were still purely academic experiments, with zero practicality. And it had been like this for half a century. Back then, even the best models couldn't tell cats from dogs. Today every high schooler with a laptop can solve that problem thanks to abundant har…
The only worthwhile thing it mentions is Ray Kurzweil, then discards it out of hand without actually considering the methodology because it must be unsound to be so precise : > The few who did predict what ended up happening, notably Ray Kurzweil, made lots of other confident predictions (e.g., the Singularity around 2045) that seemed so absurdly precise as to rule out the possibility that they were using any sound m…
Kurzweil's best arguments are really quite simple. If you plot computer per dollar on a log graph it forms a fairly smooth line like:
https://lh6.googleusercontent.com/SPIkvg3D8tlFQvQJ8OEjFpTRdV...
And you can get out pen and project it into the future. The line is actually a bit faster increasing than exponential so curves up.
You can then compare that to human capabilities to get rough dates for stuff https://www.researchgate.net/figure/Kurzweils-8-71-chart-of-...
and that's pretty much all there is to it. It's held up quite well so far and has a good chance of doing so in the future.
It's not really that the increases were slow in the old days but that your old pc going faster didn't really mean much to people whereas overtaking human capabilities may be more of a thing.
Re: On being wrong about AI
#77The unintuitive thing about self driving cars is when we drive it’s not just about the source, destination, road lights and signs. That may be the 99% but the last 1% is all sorts of crap on the road that one has to triage in milliseconds whether it’s safe or not to run over.
Pedestrians slight gestures, animals on the road, construction, police, wet cement, pot holes, fallen trees etc.
One needs to have somewhat a human experience and know the physics and behavior of almost all objects, including infering and learning on the fly unknown objects and environments.
Now if you build a system that does that, not only have you solved self driving cars, but robotics itself.
And if you solved robotics, that is inches away from AGI. Only a matter of time before it’s learned all physical trades, and only a matter of time a few humans use an army of robots to take over, or the robots themselves doing it in the goal of self preservation.
So inventing AGI is kind of conditional to L5 self driving and once you solve AGI, it’s a very different unpredictable world.
I fundamentally believe alignment is impossible. Sure you can align AGI to a few powerful humans but something aligned to all humans is very unlikely going to happen.
Re: On being wrong about AI
#78Earlier quoted context omitted.
To be fair, none of those are ultimately a matter of "can't" so much as "don't really want to." When we got a billionaire who wanted serious space colonization, we got SpaceX, which is currently working on a moon lander that qualifies as a base in its own right. And of course, we had supersonic jets, there just wasn't enough demand. People want cheap flights more than super-fast flights. We often get excited about th…
Or it might be one way in which it's exactly the same.
I think you're underestimating the hype potential of "it's an app that I can have a normal conversation with."
Re: On being wrong about AI
#79Today's AI is simply a great way of indexing already-known information and making it readily accessible via a prompt or an API. That's extremely useful and will yield new levels of productivity, but it's not what we generally think of as being "AI."