I'd prefer we stop improving AI before we hit ASI. But if we don't, I think we should implement something like socialism right before.Â
I am not currently a socialist and have mostly identified in the abundance center left wing of liberalism. Socialism is a loaded term so let me try to clarify. What I mean is stripping all humans of longterm control of capital and ceding it to the state right before we hit ASI (loosely defined), and then ceding control to that ASI with a equal distribution constitution. This ASI may or may not use markets or other economic tools, but it's goal will just be to give us good, equalish lives. If it needs to run sub agent pseudo capitalism it will do that, but we need not be involved. Again, this isn't what I want but I'd rather have this than let the capital owners own galaxies and control politics because they made or inherited a bunch of money right before the singularity.Â
One analogy is gorillas in a zoo. The gorillas survive because we give them food. However we fix the amount of food they will receive and otherwise ~ let them do their thing. The gorillas then live in some sort of semi-natural tribal type thing.Â
Do we say the gorillas are capitalists, since they are funded by capitalist society?
No, they are gorillas in a zoo.Â
Do we give the strongest or sexiest gorilla more bananas?
No, that makes no sense if we care about gorilla utility.Â
If we get ASI it will just do whatever it wants or whatever it thinks is right, so why should we seize the means of production first?
The first ASI might be corrigible, in which case the owners of it actually will be able to point it at their goals. Even if it isn't, there might be some small chance that it isn't worth it taking control of earth as opposed to just setting off into the solar system or working with us. We should make sure in that case it is working with the people and not some small few.Â
What about thorny issues of inelastic supply like who gets to live in New York City?
There will be many thorny issues but I'm sure the ASI can think of something relatively fair and utility promoting.
Why is this different than all the other times we tried socialism?
Usually there are two arguments against socialism, bad political incentives and bad economic incentives. In terms of the internal politics and corruption I'm not sure. I don't really have strong thoughts on the ASI political economy. It might be easy enough to have a benevolent singleton or maybe it's impossible. But in terms of loss of efficiency this time it is different. The economic argument for capitalism is that it boosts productivity. If singleton socialism is 1% as efficient as post ASI capitalism, that will still have far more than enough production to give us all extremely reasonable standards of living.Â
Isn't this plan incredibly half baked?
Yes, but so is our current trajectory.Â
Â
what you are describing is not like any socialsm that has been defended. this sounds much more like enlightened despotism by an incontestable ruler
Yea that's fair I was pretty loose with the meaning of words. I think my general point is once we have ASI it seems at best unclear why we should let some humans have more control than others based on their pre ASI capital.Â
i see your point, but i am wary both of authoritarianism (even if we don't call it that) and arguments of the form "problem x exists because we don't have enough information, ASI will definitionally have enough information, so if ASI, problem x dissolves." The former is bad for myriad reasons, and the latter either doesn't seem to have much to say or bakes in controversial assumptions about problem x (in this case, that just economic distribution is a solvable and mere epistemic problem). I have more to write on this conjunction if you are interested. I aim to publish by the end of the month.Â
I'm also super wary of this, but I think these words are super complicated... I'm not sure why you think my (half-baked, not well defined) suggestion is more "authoritarian" than what would result from e.g. no significant policy changes between now and ASI (not a fair comparison but just to be extreme). If you jack wealth inequality and tech to it's limit then probably I won't be able to afford to live on earth anymore even if I have nearly unlimited material goods. I guess in your view this is less authoritarian than whatever I'm suggesting.Â
This seems more vague than what I am saying. I think my core claim is approximately: We can incentivize humans to do things with money. This is net good because even though human wealth has diminishing returns to utility (and/or no god given right to more power/control than others), the increased wealth creation more than offsets the inefficient distribution of consumption and unfair distribution of power. If we get 1000000x wealthier we should shift more and more towards re distribution (edit: additionally relationship between redistribution and incentives might break down in the case of ASI but less sure). However this seems improbable to instantiate after ASI so we should do it before.Â
Theoretically you could make a similar argument without ASI. "Once we have figured out all tech and optimized energy/wealth creation, there is little reason to incentivize innovation".Â
Writing it out it the whole argument is definitely quite leaky, hopefully whatever you publish can be more sound.Â
Â