4 ms·

Valuing Humans in the Age of Superintelligence: HumaneRank

10 points by jlaporte 2y ago

jlaporte 2y ago
"A freedom-preserving proposal for distributing wealth in a post-AI society"
PaulHoule 2y ago
Might be better to have a hubs and authority score. It could be “being a good producer” and “being a good judge” are different things. Problem is you can clone somebody’s hub score by stealing their judgements.
- jlaporte 2y ago
  > It could be “being a good producer” and “being a good judge” are different things Yes I agree with this in principle. Although in this case what "good" is, is ill-defined - or rather different to everyone. The scoring each person performs is actually most directly a statement of utility to the scorer. That should probably go into the piece as an edit :-)
- jlaporte 2y ago
  btw I also agree that pure PageRank on the endorsement links might benefit from tweaking, although some of the common failure cases for PR on the web don't apply here - you can't generate piles of extra humans to endorse-link to yourself.
  - AstralStorm 2y ago
    However you can convince or bribe humans to give you an extra rating. Ever were in a "democracy" of today? The winners have money and/or backing. Any rating thst isn't bribe proof or mutual ranking proof will get instantly gamed and you end up in a neofeudalism.
    - jlaporte 2y ago
      addressed in the piece
- johnnyo 2y ago
  Also consider that a good judge is doing a useful service to society, but isnt necessarily in the business of maximizing other people’s happiness.
terminalbraid 2y ago
Maybe let's not build the Torment Nexus.
- sebastiennight 2y ago
  Great, you just had to give them the idea of adding "humane" punishment to disincentivize low Rank behavior...
causal 2y ago
A popularity-based wealth distribution scheme. What could go wrong? Reminds me of https://community-sitcom.fandom.com/wiki/MeowMeowBeenz https://community-sitcom.fandom.com/wiki/MeowMeowBeenz
- jlaporte 2y ago
  So, to respond to the snark seriously, the problem it is trying to solve assumes that we have encountered a particular economic failure case - that AI has outcompeted humans in cost and quality of work, such that most humans have zero economic value. So without some scheme, there will be no wealth distribution whatsoever, and potentially grim actions by national leadership. It's a "utility-based" wealth distribution scheme. I prefer to find a distribution scheme that is not isomorphic to communism.
  - ben_w 2y ago
    > I prefer to find a distribution scheme that is not isomorphic to communism. Communism as it was practiced, or communism as Marx envisioned it? Because what Marx was envisioning was communal ownership of the means of production, which in the case of AI would mean communal ownership over the very businesses that necessarily have to replace humans with AI for cold hard business reasons. I don't think Marx's idea of utopia is stable, game theoretically, but then very little utopian ever is. What you've got certainly has enough merit to be interesting! But, as Communism demonstrated, the difference between theory and practice can have a genocide or three between them.
    - jlaporte 2y ago
      Either one :-) I'm not inclined to run the communism experiment again, I think we've got enough data. "20th Century Democide" https://www.hawaii.edu/powerkills/DBG.TAB1.2.GIF https://www.hawaii.edu/powerkills/DBG.TAB1.2.GIF from: https://www.hawaii.edu/powerkills/20TH.HTM https://www.hawaii.edu/powerkills/20TH.HTM
      - AstralStorm 2y ago
FridgeSeal 2y ago
Calls it “HumaneRank”, this is already off to a shaky start. > Imagine a system in which every month, every human in the nation state or other political polity participates in an exercise of the grant of endorsements to other humans My guy, we’ve literally written dystopian sci-fi stories about this. The most 1:1 identical one being that Black Mirror episode. How can you possibly think this is a good idea?
- jlaporte 2y ago
  Read the piece?
ben_w 2y ago
Interesting idea, I bet it can still be gamed because this is fancy voting and voting gets gamed a lot — the devil is in the details, rather than the grand vision.
johnnyo 2y ago
One major point I don’t see addressed, are endorsements public or private? I could see major (but different) issues with both scenarios.
FireSquid2006 2y ago
Regardless of if you're "right" or not, I cannot imagine living in this society as an individual to br anything but depressing.
m3kw9 2y ago
If superintelligence means a robot that looks, acts like a human in every way plus infinitely smarter, that does not mean human is lower value. There are people smarter and richer than me, but so what? Maybe they got lucky, and had better upbringing etc. same with machines/agi
- AstralStorm 2y ago
  There are three assumptions here: 1) that the AI is actually smarter - not the case 2) that the robot is total higher efficiency - not the case, due to expensive resources and high density fuel 3) that the human is not degraded by the economic system below the pure machine - also not the case The current winning mode is one of a cyborg - a human empowered with powerful tools, which may include robots and AI. And one human can only control so much.
yesbut 2y ago
This is terrible idea that will be aggressively rejected by humanity. This is exactly the type of "tech bros are completely out if touch with how people work" vision of the future I guess we should expect from the guy that wrote this article. Humanity has never lived in a world where leisure and idle time are not constantly being invaded by the oppressive survival demands or a hierarchal authoritarian mandate from a ruling class. If AI does provide a mechanism to allow humans to spend their free time in true leisure, like most of humanity living in the Star Trek future, then we should embrace it. Not straddle humans with artificial subordination. Reject this as all costs.
- jlaporte 2y ago
  Hey yesbut The possibility of "humans to spend their free time in true leisure" is actually cited and specifically not rejected in the piece. It's addressing a very specific, and very dangerous, failure mode of a post-AI society. Read the piece and be enlightened. Also, jeez, I pine for the day when silly randomly-applied ad-hominems like "tech bros" just get dogpiled. For your own sake, elevate your discourse, man (or woman).
  - yesbut 2y ago
    no
tristor 2y ago
Isn't this the plot to the episode "Nosedive" of Black Mirror? Isn't the outcome of such a system obvious? It's very clear that allowing people to rank one another in a way that results in dire economic outcomes results in a deeply bifurcated society, and in this scenario there's no out.
- AstralStorm 2y ago
  On the upside, the ones who opt into the system would likely be mentally destroyed. This includes the popular ones due to having to be slaves to popularity.

10 points by jlaporte 2y ago

"A freedom-preserving proposal for distributing wealth in a post-AI society"

Might be better to have a hubs and authority score. It could be “being a good producer” and “being a good judge” are different things. Problem is you can clone somebody’s hub score by stealing their judgements.

jlaporte 2y ago
> It could be “being a good producer” and “being a good judge” are different things Yes I agree with this in principle. Although in this case what "good" is, is ill-defined - or rather different to everyone. The scoring each person performs is actually most directly a statement of utility to the scorer. That should probably go into the piece as an edit :-)
jlaporte 2y ago
btw I also agree that pure PageRank on the endorsement links might benefit from tweaking, although some of the common failure cases for PR on the web don't apply here - you can't generate piles of extra humans to endorse-link to yourself.
- AstralStorm 2y ago
  However you can convince or bribe humans to give you an extra rating. Ever were in a "democracy" of today? The winners have money and/or backing. Any rating thst isn't bribe proof or mutual ranking proof will get instantly gamed and you end up in a neofeudalism.
  - jlaporte 2y ago
    addressed in the piece
johnnyo 2y ago
Also consider that a good judge is doing a useful service to society, but isnt necessarily in the business of maximizing other people’s happiness.

terminalbraid 2y ago

Maybe let's not build the Torment Nexus.

sebastiennight 2y ago
Great, you just had to give them the idea of adding "humane" punishment to disincentivize low Rank behavior...

causal 2y ago

A popularity-based wealth distribution scheme. What could go wrong? Reminds me of https://community-sitcom.fandom.com/wiki/MeowMeowBeenz https://community-sitcom.fandom.com/wiki/MeowMeowBeenz

jlaporte 2y ago
So, to respond to the snark seriously, the problem it is trying to solve assumes that we have encountered a particular economic failure case - that AI has outcompeted humans in cost and quality of work, such that most humans have zero economic value. So without some scheme, there will be no wealth distribution whatsoever, and potentially grim actions by national leadership. It's a "utility-based" wealth distribution scheme. I prefer to find a distribution scheme that is not isomorphic to communism.
- ben_w 2y ago
  > I prefer to find a distribution scheme that is not isomorphic to communism. Communism as it was practiced, or communism as Marx envisioned it? Because what Marx was envisioning was communal ownership of the means of production, which in the case of AI would mean communal ownership over the very businesses that necessarily have to replace humans with AI for cold hard business reasons. I don't think Marx's idea of utopia is stable, game theoretically, but then very little utopian ever is. What you've got certainly has enough merit to be interesting! But, as Communism demonstrated, the difference between theory and practice can have a genocide or three between them.
  - jlaporte 2y ago
    Either one :-) I'm not inclined to run the communism experiment again, I think we've got enough data. "20th Century Democide" https://www.hawaii.edu/powerkills/DBG.TAB1.2.GIF https://www.hawaii.edu/powerkills/DBG.TAB1.2.GIF from: https://www.hawaii.edu/powerkills/20TH.HTM https://www.hawaii.edu/powerkills/20TH.HTM
    - AstralStorm 2y ago

FridgeSeal 2y ago

Calls it “HumaneRank”, this is already off to a shaky start. > Imagine a system in which every month, every human in the nation state or other political polity participates in an exercise of the grant of endorsements to other humans My guy, we’ve literally written dystopian sci-fi stories about this. The most 1:1 identical one being that Black Mirror episode. How can you possibly think this is a good idea?

jlaporte 2y ago
Read the piece?

ben_w 2y ago

Interesting idea, I bet it can still be gamed because this is fancy voting and voting gets gamed a lot — the devil is in the details, rather than the grand vision.

johnnyo 2y ago

One major point I don’t see addressed, are endorsements public or private? I could see major (but different) issues with both scenarios.

FireSquid2006 2y ago

Regardless of if you're "right" or not, I cannot imagine living in this society as an individual to br anything but depressing.

m3kw9 2y ago

If superintelligence means a robot that looks, acts like a human in every way plus infinitely smarter, that does not mean human is lower value. There are people smarter and richer than me, but so what? Maybe they got lucky, and had better upbringing etc. same with machines/agi

AstralStorm 2y ago
There are three assumptions here: 1) that the AI is actually smarter - not the case 2) that the robot is total higher efficiency - not the case, due to expensive resources and high density fuel 3) that the human is not degraded by the economic system below the pure machine - also not the case The current winning mode is one of a cyborg - a human empowered with powerful tools, which may include robots and AI. And one human can only control so much.

yesbut 2y ago

This is terrible idea that will be aggressively rejected by humanity. This is exactly the type of "tech bros are completely out if touch with how people work" vision of the future I guess we should expect from the guy that wrote this article. Humanity has never lived in a world where leisure and idle time are not constantly being invaded by the oppressive survival demands or a hierarchal authoritarian mandate from a ruling class. If AI does provide a mechanism to allow humans to spend their free time in true leisure, like most of humanity living in the Star Trek future, then we should embrace it. Not straddle humans with artificial subordination. Reject this as all costs.

jlaporte 2y ago
Hey yesbut The possibility of "humans to spend their free time in true leisure" is actually cited and specifically not rejected in the piece. It's addressing a very specific, and very dangerous, failure mode of a post-AI society. Read the piece and be enlightened. Also, jeez, I pine for the day when silly randomly-applied ad-hominems like "tech bros" just get dogpiled. For your own sake, elevate your discourse, man (or woman).
- yesbut 2y ago
  no

tristor 2y ago

Isn't this the plot to the episode "Nosedive" of Black Mirror? Isn't the outcome of such a system obvious? It's very clear that allowing people to rank one another in a way that results in dire economic outcomes results in a deeply bifurcated society, and in this scenario there's no out.

AstralStorm 2y ago
On the upside, the ones who opt into the system would likely be mentally destroyed. This includes the popular ones due to having to be slaves to popularity.