EAs would mostly not have been abolitionists
An imagined comment exchange on EA Forum in 1858 (3 years before the start of the US Civil War):
EA Forum in 1858: “My uncle has slaves— that doesn’t mean I’m bad! You’re lumping us all together.”
Abolitionist: “Are you going to oppose him?”
EAF 1858: “He’s my uncle! You’re saying it’s bad to be in my family but I can’t help it. That’s just who I am!”
Abolitionist: “You can tell him he’s wrong to have slaves and not contribute to his business. You can be part of the abolitionist movement.”
EAF 1858: “You’re so extreme and purist. How am I supposed to have the plantation-running job I want in the abolitionist movement? That’s what I trained for. That’s what my family does.”
Abolitionist: “That entire system is predicated on enslaving people to work the fields. That’s not what an ethical future for you is ever going to look like. You have a chance to free people in chains— there’s so much more honor in that.”
EAF 1858: “Maybe I would have helped you if you didn’t make me feel bad, but since you didn’t follow my plantation etiquette, in which I am a good person (and in which the right to own people is baked in), I’m going to smugly stay here.”
More here: https://x.com/ilex_ulmus/status/2003892377220452412?s=20
I find this pretty unsavory. I identify as EA, but not particularly involved in the AI safety space, as I am more focused on animal welfare.
I have not been shy about drawing the comparison between slavery abolition and factory farming abolition, so I am someone who is willing to play that card so to speak.
Still, it seems to me that reasonable minds can disagree about the importance of AI safety and p(doom). AGI is hardly a surefire thing. The atrocities of slavery (and factory farming) are much realer and more tangible. I think comparing something that is not inevitable to that amount of visceral suffering, that suffering that has actually occurred doesn’t feel constructive.
I understand you are writing this to get a rise out of people just like me. But this misses the mark in my personal opinion.
Ah so your objection is that the issue is not important enough to compare to slavery in this way? Interesting, wasn’t expecting that here. Are you saying that more EAs would have been abolitionists because slavery was a tangible harm? Looking around at the number of EAs who aren’t vegetarians or animal donors I’m not so sure of that.
The people who want to work with the system on AI Safety generally do believe AI risk is that important, though. Or at least they say that.
In my vocabulary, “abolitionist” means “person who is opposed to slavery” (e.g. “would vote to abolish slavery”). My sense is that this is the common meaning of the term, but let me know if you disagree.
It seems, then, that the analogy would be “person who is opposed to factory farming” (e.g. “would vote to outlaw factory farms”), instead of “vegetarian” and “animal donor”. The latter two are much higher standards, as they require personal sacrifice (in the same way that “not consuming products made with slave labor” was a much higher standard—one that very few people held themselves to).
The vast majority of EAs oppose factory farming, and I think they would have also supported the abolition of slavery.
What’s your point? Tolerating slavery would have been fine as long as they thought it was wrong in theory?
No, just that, contra your title, most EAs would have been abolitionists, the way that I understand the word “abolitionist” to be used.
Cool, let’s just redefine “alignment” and call it a day. Wait, isn’t that kind of what you’re doing?
Your hypothetical EA sounds kind of like an abolitionist to me :)
This isn’t abolitionism. Manumission means “letting go”.
“Its mandate was to promote gradual emancipation”
Yeah, freeing slaves, but not necessarily abolishing the institution. I’m not trying to be difficult— I think this difference in goals is the point.
And it’s fine if you want to bite the bullet and say you wouldn’t be a radical abolitionist, but most modern people think they would have.
What you quoted is more like voluntary safety commitments— I think both anti-slavery and AI Safety should not be left to the whims of the guilty parties to carry out.
Yeah, I agree that the hypothetical EA seems less like a “radical” abolitionist (for some definition of “radical”).