This is the most in-depth estimate I've seen of when the first AGI will be invented. Most of the expert polls and methodologies converge on the 2050 - 2070 timeframe.
https://astralcodexten.substack.com/p/b ... that-might
I agree with them.
2050 - 2070 window for inventing the first AGI
- funkervogt
- Posts: 1178
- Joined: Mon May 17, 2021 3:03 pm
Re: 2050 - 2070 window for inventing the first AGI
Move back by a decade and that's hitting the spot. 2040-2050 is more likely. We have an abysmal record of being optimistic but equally so for conservative estimates.funkervogt wrote: ↑Thu Feb 24, 2022 8:24 pm This is the most in-depth estimate I've seen of when the first AGI will be invented. Most of the expert polls and methodologies converge on the 2050 - 2070 timeframe.
https://astralcodexten.substack.com/p/b ... that-might
I agree with them.
To know is essentially the same as not knowing. The only thing that occurs is the rearrangement of atoms in your brain.
Re: 2050 - 2070 window for inventing the first AGI
I have to agree with Raklian
2050-2070 seems unusually late for AGI. Now for something like the Singularity— that is, the supremacy of AI over human intelligence— I'd definitely put that in that time frame.
But AGI does not seem to be too far beyond us. Or perhaps I'm wrong. Perhaps what I'm seeing is proto-AGI instead and there'll be a substantial gap in time before we make the leap crossing over from proto-AGI to the real thing.
2050-2070 seems unusually late for AGI. Now for something like the Singularity— that is, the supremacy of AI over human intelligence— I'd definitely put that in that time frame.
But AGI does not seem to be too far beyond us. Or perhaps I'm wrong. Perhaps what I'm seeing is proto-AGI instead and there'll be a substantial gap in time before we make the leap crossing over from proto-AGI to the real thing.
And remember my friend, future events such as these will affect you in the future
Re: 2050 - 2070 window for inventing the first AGI
Regardless of what happens.
I am just waiting untill 2030 and if we don't have it I will just set a new goal of waiting another 5 years or so.
I don't think I have the ability to wait for 2050-2070 so for me it is more so manageable long term waiting goals in chunks.
Hopefully by 2030 my life will be sorted and I will enjoy life somewhat making the next 20+ years feel easier akin to how an average person feels time.
Personally I trust Yuli's proto AGI prediction but think real AGI will come soon after it to me the skill gap doesn't seem that big in exponential growth terms and "Real AGI" includes AGI that is better than humans at AI research and nothing else as it has the same soft take off effect of adding an extra exponential to AI.
I am just waiting untill 2030 and if we don't have it I will just set a new goal of waiting another 5 years or so.
I don't think I have the ability to wait for 2050-2070 so for me it is more so manageable long term waiting goals in chunks.
Hopefully by 2030 my life will be sorted and I will enjoy life somewhat making the next 20+ years feel easier akin to how an average person feels time.
Personally I trust Yuli's proto AGI prediction but think real AGI will come soon after it to me the skill gap doesn't seem that big in exponential growth terms and "Real AGI" includes AGI that is better than humans at AI research and nothing else as it has the same soft take off effect of adding an extra exponential to AI.
Re: 2050 - 2070 window for inventing the first AGI
Now, 3 months later, are you still in the 2050 - 2070 window ?
Re: 2050 - 2070 window for inventing the first AGI
That is a really good question, I too am curious to see how to advances made in just the last 3 months (feb-may), that some might argue are more impressive than the entire last year, have impacted your estimations for AGI.
- funkervogt
- Posts: 1178
- Joined: Mon May 17, 2021 3:03 pm