I recently read someone’s call to action on AI safety[1] that included a line along the lines of “we need someone to convince your parents”, and it really stuck with me.
Do we need someone to convince my parents? Who is “we”, and why is it important for my parents to know anything about AI safety?
This also sounds kind of familiar. There is another movement I can think of that sometimes uses messaging like this. This is not a new observation by any means, but to me, it feels like the AI safety movement is increasingly “becoming” the environmental movement in a variety of ways.
Let me say upfront: I’m not saying this is necessarily a bad thing, at least on consequentialist grounds. The environmental movement has achieved a lot: it has improved the lives of many people, animals, and it has protected the environment! But this trend does make me a little worried and sad, because I think the environmentalist movement has made a number of mistakes, and it would be bad for the AI safety movement to repeat them.
In what ways is the AI safety movement becoming like the environmental movement? Here are some of my observations.
Arguments are now soldiers. Eliezer wrote: “Once you know which side you're on, you must support all arguments of that side, and attack all arguments that appear to favor the enemy side; otherwise it's like stabbing your soldiers in the back - providing aid and comfort to the enemy. People who would be level-headed about evenhandedly weighing all sides of an issue in their professional life as scientists, can suddenly turn into slogan-chanting zombies when there's a Blue or Green position on an issue.” Scott Alexander wrote that he felt he was retaliated against (admittedly in a perhaps minor way) by an organization in the space for believing the “wrong” arguments.
When people in the movement talk about climate or AI risk, they operate in a frame where there is a consensus, right answer. Anyone that disagrees must be either dumb or evil. Some people in the AI safety community act as if it’s impossible for someone who is smart and has seriously engaged with the arguments to come away disagreeing with them. That is a concerning thing to believe!
“But there is an actual consensus, at least in the case of climate science,” you say. (Don’t worry, I am not a climate change denier.) However, I notice that the label “climate change denier” is liberally applied to people who do agree with the scientific consensus (that there is human-caused climate change via CO2 emissions) but who hold other, less accepted views (perhaps that the impact of climate change won’t be that severe, or perhaps that we should solve the problem via different policies.) To be clear, I don’t think this is happening in AI safety. But in climate, it’s very common to label someone a “climate change denier,” which implies that they are no long worth listening to or engaging with. At my last job, I worked on AI for weather forecasting at a climate-tech startup. I follow a meteorologist on Twitter who is known as something of a climate skeptic but who also posts interesting content on meteorology. Every time I crossposted something of his, it was hard to get people to judge it on its own merits before someone went “oh, it’s from that climate change denier” and everyone stopped looking.
Anything that happens is interpreted as evidence for doom, or as evidence that the situation is worse than we thought.
I opened Twitter to find an example of this and I thought, hmmm, I bet David Wallace-Wells does this sort of thing. And there it was: one of the latest tweets on his timeline was a quote tweet of a story about trees in the UK experiencing spring six months early.
On the AI safety side, I’m starting to see this happen more and more. Rob Wiblin’s tweet about the recent cyberattacks on a French supermarket is just one example that comes to mind.
To be clear, in a lot of these cases, I think the posts are correct on the object level. As far as I could tell, climate change really is causally responsible for what’s happening to the trees in the UK, and the French supermarket cyberattacks probably do foreshadow how serious future AI risks could be. However, it seems that each time, people’s reactions are not simply independent, level-headed assessments of the situation, but also an effort to help paint the “doom” narrative—a collective ritual of confirmation bias. Of course, it is well known that the media environment incentivizes this behavior for everyone, and it isn’t really fair to label this as a specialty of the environmental or AI safety movements.
Appeal to emotions and scare tactics. Think of the classic climate documentary move of showing animals in distress, or “the earth is weeping.” The first chapter of Silent Spring is Rachel Carson’s fable of a perfect town that gets destroyed by an “evil spell”. And here are the modern day examples for both environmental and AI safety:
(Sorry David and Rob, I just happened to be on your Twitter profiles and you supply good examples.)
Lastly, the part of the AI safety movement that opposes improving prosaic alignment reminds me a bit of the part of the environmentalist movement that opposes building clean energy. Again, there are obvious disanalogies, but one lens that explains both is that solutions like these cannot be the right ones, even if they are effective, because they don’t further the real goal—stopping people from building superintelligent AI or dismantling capitalism.
Why is this happening? Perhaps these things just arise naturally as any movement gets bigger. And perhaps the default hypothesis should be that both movenents are correct to engage people this way. Maybe this is what it takes to win in politics, and both movements have goals that can only be achieved through policy. I am totally open to this being the case.
Nonetheless, I’m slightly concerned about these observations.
In Confessions of a Greenpeace Dropout, Patrick Moore, ecologist and former president of Greenpeace Canada says:
To a considerable extent the environmental movement was hijacked by political and social activists who learned to use green language to cloak agendas that had more to do with anticapitalism and antiglobalization than with science or ecology.
(By the way, when Claude reviewed an earlier draft of this essay, they flagged that Patrick Moore is a “climate change denier” and citing him might discredit this essay for some readers. Make what you will of this.)
Jerusalem Demsas and Matt Yglesias (hopefully less controversial figures) make a similar point on an episode of The Argument podcast. Essentially, the climate movement has been captured by anti-growth sentiment and lost sight of its ultimate goal: defending human flourishing against the risk of environmental destruction. Instead of pursuing clean energy technology and effective policy, the movement has shifted its focus to attention-grabbing protests and to blocking the buildout of clean energy.
Among today's young people, being bought into the climate cause is very much the default. I think part of the environmentalist movement has done incredible damage with its influence. Degrowth wouldn't be even remotely mainstream if not for the environmental movement's success. I remember a conversation with a friend in college about whether to have kids someday, and I cited concerns that it would be bad for the environment!
The more concrete harms I can point to are the opposition to nuclear power and the opposition to GMOs. One economic model estimates that delays in approving golden rice, driven largely by anti-GMO opposition in which Greenpeace played a leading role, cost India about 1.4 million DALYs between roughly 2002 and 2012. On the nuclear side, Jarvis et al. (2022) estimate that Germany’s shutdown of nuclear power plants post-Fukishima, which were replaced largely by coal and imports, cost about $12 billion per year.
What is AI safety’s equivalence of banning nuclear power, or of anti-GMO activism? What is AI safety’s paper straw? I would really, really like for us to to avoid these.
By the way, none of these observations should be interpreted as reasons to think concerns about existential risk from AI are not valid and real. And from what I know about the AI safety movement, and with the history of the environtal movement available to study, I have fundamentally higher hopes that we will be able to avoid falling into the same traps. I hope this essay adds a small piece to the puzzle and helps us navigate this situation more thoughtfully and wisely.
It’s from Celeste’s piece which I recently wrote an essay criticizing. It seems that Celeste’s Substack has since become subscriber only. But this sentence was quoted in this comment on my essay. Thank you, Celeste, for inspiring many thoughts in me that I wouldn’t have thought otherwise!
I recently read someone’s call to action on AI safety[1] that included a line along the lines of “we need someone to convince your parents”, and it really stuck with me.
Do we need someone to convince my parents? Who is “we”, and why is it important for my parents to know anything about AI safety?
This also sounds kind of familiar. There is another movement I can think of that sometimes uses messaging like this. This is not a new observation by any means, but to me, it feels like the AI safety movement is increasingly “becoming” the environmental movement in a variety of ways.
Let me say upfront: I’m not saying this is necessarily a bad thing, at least on consequentialist grounds. The environmental movement has achieved a lot: it has improved the lives of many people, animals, and it has protected the environment! But this trend does make me a little worried and sad, because I think the environmentalist movement has made a number of mistakes, and it would be bad for the AI safety movement to repeat them.
In what ways is the AI safety movement becoming like the environmental movement? Here are some of my observations.
At the top of David Wallace-Wells’s Twitter profile.
Rob Wiblin tweet
(Sorry David and Rob, I just happened to be on your Twitter profiles and you supply good examples.)
Why is this happening? Perhaps these things just arise naturally as any movement gets bigger. And perhaps the default hypothesis should be that both movenents are correct to engage people this way. Maybe this is what it takes to win in politics, and both movements have goals that can only be achieved through policy. I am totally open to this being the case.
Nonetheless, I’m slightly concerned about these observations.
In Confessions of a Greenpeace Dropout, Patrick Moore, ecologist and former president of Greenpeace Canada says:
(By the way, when Claude reviewed an earlier draft of this essay, they flagged that Patrick Moore is a “climate change denier” and citing him might discredit this essay for some readers. Make what you will of this.)
Jerusalem Demsas and Matt Yglesias (hopefully less controversial figures) make a similar point on an episode of The Argument podcast. Essentially, the climate movement has been captured by anti-growth sentiment and lost sight of its ultimate goal: defending human flourishing against the risk of environmental destruction. Instead of pursuing clean energy technology and effective policy, the movement has shifted its focus to attention-grabbing protests and to blocking the buildout of clean energy.
Among today's young people, being bought into the climate cause is very much the default. I think part of the environmentalist movement has done incredible damage with its influence. Degrowth wouldn't be even remotely mainstream if not for the environmental movement's success. I remember a conversation with a friend in college about whether to have kids someday, and I cited concerns that it would be bad for the environment!
The more concrete harms I can point to are the opposition to nuclear power and the opposition to GMOs. One economic model estimates that delays in approving golden rice, driven largely by anti-GMO opposition in which Greenpeace played a leading role, cost India about 1.4 million DALYs between roughly 2002 and 2012. On the nuclear side, Jarvis et al. (2022) estimate that Germany’s shutdown of nuclear power plants post-Fukishima, which were replaced largely by coal and imports, cost about $12 billion per year.
What is AI safety’s equivalence of banning nuclear power, or of anti-GMO activism? What is AI safety’s paper straw? I would really, really like for us to to avoid these.
By the way, none of these observations should be interpreted as reasons to think concerns about existential risk from AI are not valid and real. And from what I know about the AI safety movement, and with the history of the environtal movement available to study, I have fundamentally higher hopes that we will be able to avoid falling into the same traps. I hope this essay adds a small piece to the puzzle and helps us navigate this situation more thoughtfully and wisely.
It’s from Celeste’s piece which I recently wrote an essay criticizing. It seems that Celeste’s Substack has since become subscriber only. But this sentence was quoted in this comment on my essay. Thank you, Celeste, for inspiring many thoughts in me that I wouldn’t have thought otherwise!