Have we considered NOT taking that risk? Given it is extremely unlikely that we will do the necessary preparation and take precautions. It’s fine to identify issues and benefits, but then you have to ask yourself why accept that gamble in behalf of the whole world
How would that work? Make a law that it is illegal to own more than X GFLOPS of computing per person? With another limit on corporations? Maybe a Computing Enforcement Agency to investigate potential violations? On a slightly different topic, why have I never heard about taxes for AI/robots? Is there a reason why people should be the only taxpayers?
We live in a big modern world there some options that are not really available. As if we are stuck in a really wide raging river. We are not going to be going upstream regardless our desires. Saying stop I want out is not really an option.
Of course it is. We do that for lots of technologies that are considered too risky, including weapons, cryptographic algorithms, substances, and way more
I don’t really understand this. Sometimes I see even stronger versions of this sentiment, that amount to saying, hey, it’s inevitable, so you might as well be e/acc. But you can just not do things. Societies can just not do things. They used to know this. We don’t have to admit the neoliberal Nuremberg defense of “I was just following incentives.”
I really dislike the vision of the future the AI industry wants. There are so many things that could be built, why are we building potentially horrible ones?
He missed the most basic risk: unequal access. AI providers have enormous investments to pay off and enormous operating costs and want a ROI so they are going charge all the market will bear. Imagine a nightmare scenario where access to AI is like access to medical insurance coverage in the United States: white collar professionals get access to more tokens on better AI models through their employer and can afford the same for their personal endeavors. Meanwhile the working poor and unemployed are expected to make do with low cost AI models and smaller token allotments that they are grudgingly given by the tech megacorps at the cost of exposure to advertising or some other unwelcome intrusion. One could conceivably see the destitute pawning off possessions to buy enough high end LLM tokens to clean up their resume for an important job interview only to tragically blow them on their AI GF/BF because they needed a moment of reassurance in an uncaring world.
Somebody ring up Cory Doctorow; I'm sure he could wring a cautionary dystopian tale out of the concept.
https://news.ycombinator.com/item?id=49450672, meet https://news.ycombinator.com/item?id=49451313 reply
Have we considered NOT taking that risk? Given it is extremely unlikely that we will do the necessary preparation and take precautions. It’s fine to identify issues and benefits, but then you have to ask yourself why accept that gamble in behalf of the whole world
How would that work? Make a law that it is illegal to own more than X GFLOPS of computing per person? With another limit on corporations? Maybe a Computing Enforcement Agency to investigate potential violations? On a slightly different topic, why have I never heard about taxes for AI/robots? Is there a reason why people should be the only taxpayers?
We live in a big modern world there some options that are not really available. As if we are stuck in a really wide raging river. We are not going to be going upstream regardless our desires. Saying stop I want out is not really an option.
Of course it is. We do that for lots of technologies that are considered too risky, including weapons, cryptographic algorithms, substances, and way more
I don’t really understand this. Sometimes I see even stronger versions of this sentiment, that amount to saying, hey, it’s inevitable, so you might as well be e/acc. But you can just not do things. Societies can just not do things. They used to know this. We don’t have to admit the neoliberal Nuremberg defense of “I was just following incentives.”
https://en.wikipedia.org/wiki/Wicked_problem
I really dislike the vision of the future the AI industry wants. There are so many things that could be built, why are we building potentially horrible ones?
He missed the most basic risk: unequal access. AI providers have enormous investments to pay off and enormous operating costs and want a ROI so they are going charge all the market will bear. Imagine a nightmare scenario where access to AI is like access to medical insurance coverage in the United States: white collar professionals get access to more tokens on better AI models through their employer and can afford the same for their personal endeavors. Meanwhile the working poor and unemployed are expected to make do with low cost AI models and smaller token allotments that they are grudgingly given by the tech megacorps at the cost of exposure to advertising or some other unwelcome intrusion. One could conceivably see the destitute pawning off possessions to buy enough high end LLM tokens to clean up their resume for an important job interview only to tragically blow them on their AI GF/BF because they needed a moment of reassurance in an uncaring world.
Somebody ring up Cory Doctorow; I'm sure he could wring a cautionary dystopian tale out of the concept.
Dunno what's going on with these urls;
[dupe] Discussion: https://news.ycombinator.com/item?id=49448137
Weird. I searched for it and didn't see anything. My bad.
[flagged]