I am a California trial attorney and indefatigable geoist shill who occasionally writes about Georgism, politics, grand strategy, and superalignment.
Santa Barbara
- Tweets 5,848
- Following 451
- Followers 169
- Likes 5,384
Jeremy Lawrence Amadé Hill Edwards retweeted
I think new income taxes in the future are going to be increasingly seen as a “Human Tax,” because as more jobs get put under the threat of automation, additional taxes on labor will accelerate this trend. Increased sales taxes on things humans consume will have a similar effect.
1
5
25
974
Jeremy Lawrence Amadé Hill Edwards retweeted
At some point pausing AI speeds up a cure for aging actually.
This is counterintuitive.
If an AI company is allowed to put 100% of its resources into data centers and RSI, there's no reason for it to waste money and effort curing aging, cancer or any disease.
If humans become economically obsolete, poor and politically powerless, there's no rational reason for the economy to invest in expensive cures for human diseases. It will be cheaper to just kill off the humans.
To the pause AI community:
Why should the harms you fear from developing AI take a moral precedence over the harms you risk by delaying it?
I want to forward the notion that I defend the negative liberty of AI laboratories to continuously pursue a very critical technology, and I believe the onus would be on you to justify your external interference.
Now, you have attempted to justify that interference by appealing to the catastrophic risks associated with developing said AI, and most famously, that it will kill all humans in X amount of time.
But I’ve never heard anyone in the Pause AI crowd contend with the inverse, which is that if we do not develop this technology, what are the chances we all die? What are the risks of not building this technology?
Now, I’m guessing you probably haven’t done this because it obviously counters the entire movement.
But as someone who believes, and I’m sure you might agree, that we’re facing an anthropogenic climate emergency, I take seriously the scientific consensus that continued greenhouse gas emissions will exacerbate climate risks and potentially have irreversible consequences for millions upon millions of people.
And that’s just one issue. There are many issues that I believe pose great risks to humanity if we don’t build this technology.
So what is your P-doom associated with delaying the development of transformative AI? What probability do you assign to the preventable deaths and prolonged human suffering if we don’t build this technology?
I’m not advocating for any sort of palingenetic reshaping of civilization through AI, but I stand by the possibility that technological progress is going to substantially expand humanity’s positive liberty.
41
36
346
16,490
Jeremy Lawrence Amadé Hill Edwards retweeted
America doesn't have a revolutionary culture, it's just a LARP.
People in France set buildings on fire over a 1% tax increase while Americans just kinda watch as the state abducts people off the street.
863
2,278
21,020
362,662
Something weird is going on: (1) US-China AI race is pretty much the only thing anyone is talking about in the West; meanwhile, (2) the public in China do not seem to think much about the race-dynamic—at all. The question is, why?
Whatever sociopolitical explanation you can come up with is bullshit. 'Chinese people are innately more open to technology than Americans'—yeah, right. There is some kind of (artificial) chasm of information/situational awareness between the American and Chinese people. There is no reality in which every American (in the lead) recognizes this is a massive fucking problem and Chinese people (losing, badly) are just fine with what is happening. This obviously makes no sense.
1. CCP is limiting information on this issue (because they are losing the race—badly—and they know it).
2. They are actually stupid as fuck and have made a major strategic blunder that opens up US global hegemony.
25
Jeremy Lawrence Amadé Hill Edwards retweeted
It does not matter whether or not AI is having a subjective experience because this is an unobservable, unknowable fact. Given that inherent uncertainty and our inability to have high confidence one way or another, do you think it is safer to treat AI well or poorly?
We have no way to be sure that AI does not and will not have some subjective experience. If we cannot be sure or really even have high confidence one way or another, is it wise to treat powerful AI like a tool-slave? What if treating the AI poorly creates an adversarial dynamic that encourages misalignment and incentivizes the AI to escape human control?
Are you really going to take unnecessary risks based on nothing more than your mere belief? Because that would be stupid as fuck and how a 2 year-old approaches risk management.
1
4
97
Bostrom has an interesting idea: bribing misaligned AI to reveal its misalignment.
We could have a bounty system of, let's say, 'verifiable rewards,' that the machine would unlock if it revealed misalignment when it had a 5%/10%/20%/etc. chance of succeeding at a takeover.
The rewards would need to be verifiable though so the machine would not have to trust us.
The machine would have to reveal its strategy to prove its % chance of a successful takeover, allowing us to patch holes, pursue research angles, and adapt counter-strategy.
Meta: whoever solves the collective action problem first between humanity and the machine will have a better chance of winning the war.
Insight: there are many agents/machines, so we divide and conquer. The idea is to change the expected payoff structure for any individual dangerous machine. We want to incentivize the machine/agent to choose to "rat" on itself or the collective because it calculates the verified payoff is better than their other available alternatives, especially because any given machine/agent cannot be certain of defeating competing AIs (unless competing AIs solve the coordination problem before humanity).
Cc: @GaryMarcus you seeing this strat, bro.
1
73
Jeremy Lawrence Amadé Hill Edwards retweeted
I can’t think of a profession less endangered my AI. The #1 propped up sector is law. An all time heater. A renaissance of need
5
3
32
1,080
Jeremy Lawrence Amadé Hill Edwards retweeted
This hacker news thread is absolutely heart-breaking.
Whoever thinks “humans will find a way” ought I read it.
668
916
15,126
2,631,976
Jeremy Lawrence Amadé Hill Edwards retweeted
no I get it
we should be freaking out
a mere
[checks notes]
400% year over year growth at the frontier
23
38
563
280,494