Untitled topic • Hacker News Board

I know Marc is skeptical of AI safety alarmism, I doubt he would think we should nuke China over AI safety. He’s also dry and and sarcastic. I bet if I investigate this it’ll turn out he’s doing a reductio ad absurdum of his opponents position and you’ve just misunderstood or didn’t have context.

Re: undefined

#7

Post by fwlr » Mon, Mar 06, 2023, 1:48 AM UTC

fwlr wrote: post #6

I know Marc is skeptical of AI safety alarmism, I doubt he would think we should nuke China over AI safety. He’s also dry and and sarcastic. I bet if I investigate this it’ll turn out he’s doing a reductio ad absurdum of his opponents position and you’ve just misunderstood or didn’t have context.

Yep, seems like that’s exactly it.

Bob Hayes, quoting Gary Marcus: "It’s time for government to consider frameworks that allow for AI research under ... ethical standards and safety, while considering a pause on the widespread public dissemination of... AI technologies." https://twitter.com/bobehayes/status/1632146399692292097

Gary Marcus: “Still haven’t heard any serious argument against this proposal.” https://twitter.com/GaryMarcus/status/1632146619645988864

Marc Andreessen: “It worked for COVID! Seriously though. If AI really is existentially dangerous, then it can't be worked on anywhere, as the risk of a lab leak is still catastrophic. The only answer is a global authoritarian dictatorship of cosmic proportions.“ https://twitter.com/pmarca/status/1632231156807786496 and https://twitter.com/pmarca/status/1632235642691411970

Loki: “There are only two answers: either everyone needs to chill-out and admit it's not really that dangerous or the PRC needs to have their computing power involuntarily reduced via any means.” https://twitter.com/LokiJulianus/status/1632237941555691522

Marc: “Yes. A lot of places in the world are going to have to be glassed.” https://twitter.com/pmarca/status/1632238233403936768

So, yeah, as suspected, it’s a reductio of the “AI safety requires an authoritarian government response” position. He’s not saying nuke China’s server farms, he’s actually warning that authoritarian government solutions for AI safety are going to end up nuking China’s server farms.

So assuming you think glassing China is bad, you’re actually agreeing with Marc, making this a sincere case of The Worst Person You Know Just Made a Great Point: https://clickhole.com/heartbreaking-the-worst-person-you-kno...

Re: undefined

#8

Post by fspeech » Mon, Mar 06, 2023, 2:35 AM UTC

fwlr wrote: post #7

fwlr wrote: post #6

I know Marc is skeptical of AI safety alarmism, I doubt he would think we should nuke China over AI safety. He’s also dry and and sarcastic. I bet if I investigate this it’ll turn out he’s doing a reductio ad absurdum of his opponents position and you’ve just misunderstood or didn’t have context.

Yep, seems like that’s exactly it. Bob Hayes, quoting Gary Marcus: "It’s time for government to consider frameworks that allow for AI research under ... ethical standards and safety, while considering a pause on the widespread public dissemination of... AI technologies." https://twitter.com/bobehayes/status/1632146399692292097 Gary Marcus: “Still haven’t heard any serious argument against this proposal.” https://twit…

"he’s actually warning that authoritarian government solutions for AI safety are going to end up nuking China’s server farms."

He's saying that is what it takes for the policy to succeed. It's obviously absurd (to initiate a nuclear attack over someone's AI policy), which leads one to question the premise of technology control.