Kudos for a very thoughtful article. I've ruminated on technology and society issues for a long time, and there's some deep and disturbing aspects of the topic. But I've been discouraged in that there's not a lot of space between, roughly, punditry for Utopia and Dystopia.
I have some slight notes. Musk doesn't just go on about "insane racist conspiracy theories". No, he's not so limited. He posts about all sorts of insane right-wing lunacy. Many ordinary people in tech are a bit annoyingly racist, but otherwise reasonable. Musk is full-on MAGA, with the entire menu of cruelty, contempt for the poor and vulnerable, xenophobia, on and on.
When you say "frowned on in populist left-wing spaces to imply that Silicon Valley is capable of producing anything new and useful at all", this is not new. The "identitarian" left has always frowned on technology. This is a topic in itself, the actual Communists were always talking about improving production (collectively, of course).
But "To even imply AI's usefulness makes it worthy of any carbon cost at all is seen as completely missing the big picture." ... well, yes, I think you might not be giving their argument its due. I don't agree with that argument on the facts, but I grasp it. No offense meant, please forgive me here, but there's times you do give the impression of the sort of pundit who says (humor) "Have you considered that toxic waste might be GOOD for you? Or at least, that poisoning some people might be a net economic benefit if it helps make products which are enjoyed by millions? We make social trade-offs all the time, anyone who doesn't realize that is a naive dunderhead. When you complain about all the pollution, I don't ever see you acknowledge the benefits of the end results."
Many, many people on the left have an entirely reasonable reaction to that style of argument, that it's trying to pull a con job on them. THIS IS RATIONAL! There's people who are paid to outright lie like this. My impression is you've absorbed a certain pattern of argument which is extremely common in certain tech circles, but it tends to lead to conflict when the weaknesses aren't understood.
Well, this is long enough already, I hope it does some good.
Yeah to be clear I agree with everything you’re saying here, main reason I didn’t add it in the article itself was moreso that I didn’t want the section to drag and assume my readers know where I’m coming from there. The relationship between the tech elite and the tech itself is pretty complicated and I’m not dismissive of the dangers, but I do worry that I meet a lot of people who don’t consider that the tech itself might be dangerous exactly because it’s useful, because considering it useful is crossing a line
"The "identitarian" left has always frowned on technology"
They also opposed EVs early on fairly strongly, for reasons that were mostly proven false with time, and now only conservatives parrot. Now they still don't like EVs, but it's just kind of a disapproving grumbling.
(1) New technology invariably seems worse than it actually is for segments of the population that do not like it. It also invariably seems better than it is for segments of the population that like it.
In short, AI, like so much else, is a tool. As you hint (I think), people in both directions imbue it with a sort of spiritual dimension. Very few tools (any?) on Earth seems to be absolutely evil or absolutely good, but in a politics of extremes, there is a tendency to label things absolutely. Of course, the tendency is not new (see (2), below).
We had a local councilman just win a primary based in part on his opposition to any AI datacenters. Not "we should study and consider," not "I am concerned about water use aspects in an area where many homes run off the local aquifer" but simply "NO MORE." I think we have one now.
(2) Running of my first comment, I was looking over claims about the radio, which appear not unlike some of the spiritual claims made about internet, ai, etc, such as:
Broadcasting bridges the cultural and social gulf which formerly separated the villager from the town dweller, and frequently also the poor from the rich and one people from another. By communicating to all in equal measure and making itself the proclaimer of the great moral forces of mankind, it creates a community in the spiritual realm which is not limited to one land, but gradually embraces the whole earth. - Wilhelm Miklas, president of Austria, published 1932 or so.
I'm sure there is philosophical-historical paper, someplace (Ellul, Ong...?) which considers this tendency to mysticism around new technology.
Excellent read; thanks for your work on a healthy epistemic environment. I’m afraid the example of media / publics / researchers latching onto a “killer fact” that takes on a life of its own generalises to most of science, unfortunately. It’s almost like we need a new mental model of the science-media-policy interface
I am entirely in agreement with you on the object-level issue of (current) AI resource use. You are fighting a good fight with truth on your side, and I appreciate your work.
That said, I don’t think you realize how much of a cultural bubble you are in, and more importantly I don’t think you understand how people who are outside of that bubble feel about people inside of that bubble.
Your writing is funded by the Effective Altruist foundation CoefficientGiving (I do appreciate you stating this openly on your website) which in turn is funded by Facebook/Meta through Dustin Moskovitz. There are also direct connections to Anthropic through Holden Karnofsky. If AI goes well, your social group is about to take over the world. This is a threat to everyone else on Earth not part of that social group and/or not attracted to that milieu.
You made a remark about being vegan and looking down with moral contempt at people who you see eating meat. I eat meat. So do 95% of Americans. Part of my culture involves deep-frying several birds worth of chicken wings and eating them topped with ranch. If stopping tech “philanthropists” from destroying American culture means preventing data centers from being built, then I will fight to stop data centers from being built. Analogous arguments could be made for other cultural issues as well. I’m not sure we’re at that point yet, but I can see possible futures like this on the horizon.
There is also the much larger problem of, “literally everyone on Earth dying,” but you seem familiar enough with that argument already.
I mean I’m not really writing to come of as an everyman, just writing from my perspective for a specific audience. The point on veganism was supposed to be more self-effacing where I’m having a goofy fantasy of self-importance, I tried to be clear that this is all happening in my head and framed the meat eater as unknowing instead of evil. I also don’t really think about AI as being about my social group. I see EA as basically trying to comment on a new industrial revolution before it happens and suggest ways it could go better, but I think if AI does go well EA and my social group will lose a lot of influence on net, for the same reason any group that had power before the industrial revolution often got shuffled around.
I think some of the opposition is about metaphysics. It's one thing to talk up "evidence based" and being "on the side of science" when it's convenient, but a whole other matter when questions that seemed safely out of reach of finding an answer forever - like what intelligence or creativity or sentience actually are, or the whole semantics vs syntax debate - could potentially be settled.
But they're barely aware of metaphysics, so I think this is mostly unconscious. They don't even remember the "science wars" between STEM and pomo in the 1990s. They also have beliefs about sentience that have deeply buried spirituality and vitalism; debate about whether AI could have an experience and they tend to surface as gish-galosh objections. Add it up, and AI is felt to be an existential threat, an unsettling attack on foundations they were barely aware were there.
I had no idea it went this far back, just that Bender et al. 2021 seemed weirdly early to the party.
Their Parrots paper has been extremely harmful - a "current harm" if ever there were one - to the discourse as a whole, not just for its factual issues and framing, but because it presents a very attractive narrative to those already primed to mistrust big tech: if you understand that LLMs are just random nonsense generators, you have the inside scoop and aren't falling for the hype like all the rubes. A lot of smart, tech-forward people went "aha, gotcha" and closed their minds to any further nuance.
I finally got round to actually reading the paper last month, and even allowing for sincere mistakes and the march of time, it's just rough on the brain - starting with that humanities thing where "We will show that X is not Y (where for 'Y' we'll use Bob's very specific definition from his 1996 paper, which is a thing you can just choose as you please)".
Kudos for a very thoughtful article. I've ruminated on technology and society issues for a long time, and there's some deep and disturbing aspects of the topic. But I've been discouraged in that there's not a lot of space between, roughly, punditry for Utopia and Dystopia.
I have some slight notes. Musk doesn't just go on about "insane racist conspiracy theories". No, he's not so limited. He posts about all sorts of insane right-wing lunacy. Many ordinary people in tech are a bit annoyingly racist, but otherwise reasonable. Musk is full-on MAGA, with the entire menu of cruelty, contempt for the poor and vulnerable, xenophobia, on and on.
When you say "frowned on in populist left-wing spaces to imply that Silicon Valley is capable of producing anything new and useful at all", this is not new. The "identitarian" left has always frowned on technology. This is a topic in itself, the actual Communists were always talking about improving production (collectively, of course).
But "To even imply AI's usefulness makes it worthy of any carbon cost at all is seen as completely missing the big picture." ... well, yes, I think you might not be giving their argument its due. I don't agree with that argument on the facts, but I grasp it. No offense meant, please forgive me here, but there's times you do give the impression of the sort of pundit who says (humor) "Have you considered that toxic waste might be GOOD for you? Or at least, that poisoning some people might be a net economic benefit if it helps make products which are enjoyed by millions? We make social trade-offs all the time, anyone who doesn't realize that is a naive dunderhead. When you complain about all the pollution, I don't ever see you acknowledge the benefits of the end results."
Many, many people on the left have an entirely reasonable reaction to that style of argument, that it's trying to pull a con job on them. THIS IS RATIONAL! There's people who are paid to outright lie like this. My impression is you've absorbed a certain pattern of argument which is extremely common in certain tech circles, but it tends to lead to conflict when the weaknesses aren't understood.
Well, this is long enough already, I hope it does some good.
Yeah to be clear I agree with everything you’re saying here, main reason I didn’t add it in the article itself was moreso that I didn’t want the section to drag and assume my readers know where I’m coming from there. The relationship between the tech elite and the tech itself is pretty complicated and I’m not dismissive of the dangers, but I do worry that I meet a lot of people who don’t consider that the tech itself might be dangerous exactly because it’s useful, because considering it useful is crossing a line
"The "identitarian" left has always frowned on technology"
They also opposed EVs early on fairly strongly, for reasons that were mostly proven false with time, and now only conservatives parrot. Now they still don't like EVs, but it's just kind of a disapproving grumbling.
Enjoying the article, just pointing out what looks like an autocorrect mistake: "Digital outlets crew, by only added about 6,000 new jobs."
A couple thoughts on your excellent article.
(1) New technology invariably seems worse than it actually is for segments of the population that do not like it. It also invariably seems better than it is for segments of the population that like it.
In short, AI, like so much else, is a tool. As you hint (I think), people in both directions imbue it with a sort of spiritual dimension. Very few tools (any?) on Earth seems to be absolutely evil or absolutely good, but in a politics of extremes, there is a tendency to label things absolutely. Of course, the tendency is not new (see (2), below).
We had a local councilman just win a primary based in part on his opposition to any AI datacenters. Not "we should study and consider," not "I am concerned about water use aspects in an area where many homes run off the local aquifer" but simply "NO MORE." I think we have one now.
(2) Running of my first comment, I was looking over claims about the radio, which appear not unlike some of the spiritual claims made about internet, ai, etc, such as:
Broadcasting bridges the cultural and social gulf which formerly separated the villager from the town dweller, and frequently also the poor from the rich and one people from another. By communicating to all in equal measure and making itself the proclaimer of the great moral forces of mankind, it creates a community in the spiritual realm which is not limited to one land, but gradually embraces the whole earth. - Wilhelm Miklas, president of Austria, published 1932 or so.
I'm sure there is philosophical-historical paper, someplace (Ellul, Ong...?) which considers this tendency to mysticism around new technology.
Excellent read; thanks for your work on a healthy epistemic environment. I’m afraid the example of media / publics / researchers latching onto a “killer fact” that takes on a life of its own generalises to most of science, unfortunately. It’s almost like we need a new mental model of the science-media-policy interface
I am entirely in agreement with you on the object-level issue of (current) AI resource use. You are fighting a good fight with truth on your side, and I appreciate your work.
That said, I don’t think you realize how much of a cultural bubble you are in, and more importantly I don’t think you understand how people who are outside of that bubble feel about people inside of that bubble.
Your writing is funded by the Effective Altruist foundation CoefficientGiving (I do appreciate you stating this openly on your website) which in turn is funded by Facebook/Meta through Dustin Moskovitz. There are also direct connections to Anthropic through Holden Karnofsky. If AI goes well, your social group is about to take over the world. This is a threat to everyone else on Earth not part of that social group and/or not attracted to that milieu.
You made a remark about being vegan and looking down with moral contempt at people who you see eating meat. I eat meat. So do 95% of Americans. Part of my culture involves deep-frying several birds worth of chicken wings and eating them topped with ranch. If stopping tech “philanthropists” from destroying American culture means preventing data centers from being built, then I will fight to stop data centers from being built. Analogous arguments could be made for other cultural issues as well. I’m not sure we’re at that point yet, but I can see possible futures like this on the horizon.
There is also the much larger problem of, “literally everyone on Earth dying,” but you seem familiar enough with that argument already.
I mean I’m not really writing to come of as an everyman, just writing from my perspective for a specific audience. The point on veganism was supposed to be more self-effacing where I’m having a goofy fantasy of self-importance, I tried to be clear that this is all happening in my head and framed the meat eater as unknowing instead of evil. I also don’t really think about AI as being about my social group. I see EA as basically trying to comment on a new industrial revolution before it happens and suggest ways it could go better, but I think if AI does go well EA and my social group will lose a lot of influence on net, for the same reason any group that had power before the industrial revolution often got shuffled around.
I think some of the opposition is about metaphysics. It's one thing to talk up "evidence based" and being "on the side of science" when it's convenient, but a whole other matter when questions that seemed safely out of reach of finding an answer forever - like what intelligence or creativity or sentience actually are, or the whole semantics vs syntax debate - could potentially be settled.
But they're barely aware of metaphysics, so I think this is mostly unconscious. They don't even remember the "science wars" between STEM and pomo in the 1990s. They also have beliefs about sentience that have deeply buried spirituality and vitalism; debate about whether AI could have an experience and they tend to surface as gish-galosh objections. Add it up, and AI is felt to be an existential threat, an unsettling attack on foundations they were barely aware were there.
I had no idea it went this far back, just that Bender et al. 2021 seemed weirdly early to the party.
Their Parrots paper has been extremely harmful - a "current harm" if ever there were one - to the discourse as a whole, not just for its factual issues and framing, but because it presents a very attractive narrative to those already primed to mistrust big tech: if you understand that LLMs are just random nonsense generators, you have the inside scoop and aren't falling for the hype like all the rubes. A lot of smart, tech-forward people went "aha, gotcha" and closed their minds to any further nuance.
I finally got round to actually reading the paper last month, and even allowing for sincere mistakes and the march of time, it's just rough on the brain - starting with that humanities thing where "We will show that X is not Y (where for 'Y' we'll use Bob's very specific definition from his 1996 paper, which is a thing you can just choose as you please)".