It would be great if you would respond - and get Arnold Kling and Tyler and Bryan Caplan to do same as well - to this “Plan A”, and in particular the Economics component of it.
In particular, why the percentage likelihood of several of their key claims about economic growth are likely way off.
Because even if they are more correct about the “billions of genius white collar knowledge workers” point - the one place the “Plan A” authors are more expert than I am, and *arguably* at least have a decent chance of being correct (even though I highly doubt they are with their aggressive timeline) - they are IMO almost certainly wrong about the timeline of availability of robots to replace human labor. And surely wrong about many of their other economic growth probabilities.
But you - and Arnold and Tyler and Bryan, and others with credibility - could do a huge service by pointing out a few of the *specific* key areas where they are most likely very wrong on the economics (and their associated probability estimates).
Specifically over the next 14 years where this plan focuses.
Because I’m actually willing to concede that if they *are* even close to correct about their estimates of economic growth in the next 7 years, then it would actually be reasonable for government to take some actions like what they suggest.
I just don’t think they are at all correct about the economics.
But that is precisely where good, credible economists could provide real value added on this subject.
I would ask them to be more specific. Alas, I would not trust those who have consistently shown to use govt in fulfilling their own agendas, self interests, and ideological ambitions to be the arbiters of what must be done about AI, in regards to economic matters.
We can’t even get a majority of these people to remove legislated real time govt monitoring from our personal vehicles.
If the guardrails needed are about having the ability to ‘kill’ AI with a switch, or the ability for anyone to have an AI kill switch on their devices, etc., then I support that.
Wonder if they even read the whole thing or just added their names. Maybe the same people who endorsed Paul Ehrlich's book "Population Bomb". I see the next doom book bestseller by Jeffrey Sachs called the "AI Plague".
From 2011-2016 there was a TV show, Person of Interest, in which an AI was used to fight crime, but turned out to be the fight between an evil & good AI. It was a lesson as to what happens when special interests control an AI for their own interest & have power behind them. So, what happens when the governments control AI for their own special interests, or limit AI to the advantage of others?
The statement is so vague as to be almost empty, except for the overtones. It does not even say (though it does suggest) that *all* “economists, policymakers, and technology leaders” must do something with respect to AI, just as it does not say (though it does suggest) that governments must take control of the process of adjustment. Just some bad vibes.
Point 3 is pointless if any of the following are false:
. People who have the power to "steer" AI are motivated by altruistic feelings.
. People who have the power to "steer" AI have the intelligence to understand what mandates and prohibitions are beneficial and which are counterproductive.
. Any mandates and prohibitions will be honored by everyone who is developing AI.
If you read the links I attached elsewhere, they have at least a semi-plausible case for positive answers to your 3 objections. In particular, the 3rd one.
I still don’t defend “taking action” because I believe their prediction of the economics is just way, way way off.
If their economic predictions are anywhere close to true, then what they are proposing isn’t necessarily crazy.
Color me cynical, but I have no faith in the ability of economists and politicians to understand the long-term consequences and implications of AI, much less build appropriate guardrails and institutions to “steer” it toward a benign and benevolent end state. Wisdom is rare, but hubris is forever and never in short supply.
A whole lot of people think that this is a primary role for government, setting the rules of the game which will prevent a dictator from getting the best AI and taking over the world. Companies are racing to create that best version so that we don't wind up under AI's thumb. For a more balanced view, see: "The AI Doc: Or How I Became an Apocaloptimist." It starts out with the predictions of severe doom and gloomers, all academics or nonprofit types. Then it pivots. An m.d. points out that we wouldn't be here without technology, and that we've handled harder challenges than this. Listed are all the kinds of stats that Johan Norberg and you and others have listed, about the drop in poverty, hunger, etc. Though it's quite balanced, the conclusion is basically: "The only day better than today is tomorrow." You can rent it for free on Peacock by signing up for a 7-day free trial and then canceling immediately. The trial lasts the full period even if you cancel. It's here: https://www.peacocktv.com/watch-online/movies/the-ai-doc-or-how-i-became-an-apocaloptimist/1a627283-f9ed-3d44-ac81-f3955b273875
Noah Smith has an interesting piece on why he didn't sign it, largely because:
Acemoglu’s big idea — basically, to put him and his friends in charge of AI development — is not a good idea.
https://www.noahpinion.blog/p/why-i-didnt-sign-the-we-must-act
My two bits.
I expect that many of them are signing because they are worried about AI existential risk, and are implicitly endorsing something like this:
https://www.astralcodexten.com/p/introducing-plan-a
https://ai-2040.com/supplements/economics-of-plan-a
It would be great if you would respond - and get Arnold Kling and Tyler and Bryan Caplan to do same as well - to this “Plan A”, and in particular the Economics component of it.
In particular, why the percentage likelihood of several of their key claims about economic growth are likely way off.
Because even if they are more correct about the “billions of genius white collar knowledge workers” point - the one place the “Plan A” authors are more expert than I am, and *arguably* at least have a decent chance of being correct (even though I highly doubt they are with their aggressive timeline) - they are IMO almost certainly wrong about the timeline of availability of robots to replace human labor. And surely wrong about many of their other economic growth probabilities.
But you - and Arnold and Tyler and Bryan, and others with credibility - could do a huge service by pointing out a few of the *specific* key areas where they are most likely very wrong on the economics (and their associated probability estimates).
Specifically over the next 14 years where this plan focuses.
Because I’m actually willing to concede that if they *are* even close to correct about their estimates of economic growth in the next 7 years, then it would actually be reasonable for government to take some actions like what they suggest.
I just don’t think they are at all correct about the economics.
But that is precisely where good, credible economists could provide real value added on this subject.
You don't think there is anything in Yudkowsky "If anybody builds it, everybody will die".
I would ask them to be more specific. Alas, I would not trust those who have consistently shown to use govt in fulfilling their own agendas, self interests, and ideological ambitions to be the arbiters of what must be done about AI, in regards to economic matters.
We can’t even get a majority of these people to remove legislated real time govt monitoring from our personal vehicles.
If the guardrails needed are about having the ability to ‘kill’ AI with a switch, or the ability for anyone to have an AI kill switch on their devices, etc., then I support that.
As if "benefits society" had an unambiguous meaning...
To be fair, the one place where it does is “avoid human existence doom”.
Which is clearly where the folks pushing this are coming from.
See my other comment for the relevant links.
The problem is that nobody knows the future, from Prometheus and his divine enemies to Ehrlich and his population bomb. See "Running out of everything" at https://lawliberty.org/book-review/running-out-of-everything/.
I wasn’t defending everything about their claims.
Only the quite narrow one that wiping out human civilization would be - unambiguously - bad.
And you are conflating “nobody knows the future” with “‘no one has better probability estimates about the future than anyone else”.
Again, as you can see elsewhere in these comments, I ain’t a big fan of the proposal.
Because I think the economic claims quite wrong (and I think the robotics timeline way off).
And so what credible economists should do is debunk their over-the-top extraordinary claims about the economics.
Which alone would justify ignoring the rest of what they are proposing.
Wonder if they even read the whole thing or just added their names. Maybe the same people who endorsed Paul Ehrlich's book "Population Bomb". I see the next doom book bestseller by Jeffrey Sachs called the "AI Plague".
The "whole thing" is very short. I quoted the whole thing.
We radicals have to stick together, David. The other side has us badly outnumbered.
https://youtu.be/ARJ8cAGm6JE?is=xGll8iGbeQVT8-G1
From 2011-2016 there was a TV show, Person of Interest, in which an AI was used to fight crime, but turned out to be the fight between an evil & good AI. It was a lesson as to what happens when special interests control an AI for their own interest & have power behind them. So, what happens when the governments control AI for their own special interests, or limit AI to the advantage of others?
The statement is so vague as to be almost empty, except for the overtones. It does not even say (though it does suggest) that *all* “economists, policymakers, and technology leaders” must do something with respect to AI, just as it does not say (though it does suggest) that governments must take control of the process of adjustment. Just some bad vibes.
Point 3 is pointless if any of the following are false:
. People who have the power to "steer" AI are motivated by altruistic feelings.
. People who have the power to "steer" AI have the intelligence to understand what mandates and prohibitions are beneficial and which are counterproductive.
. Any mandates and prohibitions will be honored by everyone who is developing AI.
I would argue that all three of these are false.
If you read the links I attached elsewhere, they have at least a semi-plausible case for positive answers to your 3 objections. In particular, the 3rd one.
I still don’t defend “taking action” because I believe their prediction of the economics is just way, way way off.
If their economic predictions are anywhere close to true, then what they are proposing isn’t necessarily crazy.
Whether achievable or not.
Something must be done! Yes, Prime Minister. Politician's logic.
All cats have four legs.
My dog has four legs.
Therefore, my dog is a cat.
Color me cynical, but I have no faith in the ability of economists and politicians to understand the long-term consequences and implications of AI, much less build appropriate guardrails and institutions to “steer” it toward a benign and benevolent end state. Wisdom is rare, but hubris is forever and never in short supply.
They don't want to understand the future. They want to lock in yesterday's status quo.
A whole lot of people think that this is a primary role for government, setting the rules of the game which will prevent a dictator from getting the best AI and taking over the world. Companies are racing to create that best version so that we don't wind up under AI's thumb. For a more balanced view, see: "The AI Doc: Or How I Became an Apocaloptimist." It starts out with the predictions of severe doom and gloomers, all academics or nonprofit types. Then it pivots. An m.d. points out that we wouldn't be here without technology, and that we've handled harder challenges than this. Listed are all the kinds of stats that Johan Norberg and you and others have listed, about the drop in poverty, hunger, etc. Though it's quite balanced, the conclusion is basically: "The only day better than today is tomorrow." You can rent it for free on Peacock by signing up for a 7-day free trial and then canceling immediately. The trial lasts the full period even if you cancel. It's here: https://www.peacocktv.com/watch-online/movies/the-ai-doc-or-how-i-became-an-apocaloptimist/1a627283-f9ed-3d44-ac81-f3955b273875
But...but, what about Ned Ludd's job???