It seems plausible that many things could be a lot better, like making factory farming obsolete. Does that mean that animals are no longer experiencing extreme suffering? What is our baseline for animal welfare?
Furthermore, while factory farms may go away by default (with the AI transition), the same can not be said about wild animal suffering (unless all wild animals go extinct, which isn't impossible, but... bad vibes).
I think this is a good thing to point out. My main reactions are: 1. I think this work was fairly low-effort and exploratory (how could we get some quick insights using a ton of AI automation), and doesn't have the rigor I think would be needed to draw hard conclusions. For example, the Truth Social data wasn't very high quality. 2. The absence of positive-about-AI content on Bluesky is more stark and statistically significant, and I'm much more confident that a better analysis would turn up that same result.
I really appreciate this post. I also watched this video and was horrified at how no attention at all was paid to wild animal suffering, and I'm really glad there are other people who had the same experience. I think this is a serious moral blindspot in our society.
I'm currently working in technical AI safety, and I have two main thoughts on this: 1) We currently don't have the ability to robustly imbue AI with ANY values, let alone values that include all animals. We need to get a lot farther with solving this technical problem (the alignment problem) before we can meaningfully take any actions which will improve the longterm future for animals. 2) The AI Safety community generally seems mostly on board with animal welfare, but it's not a significant priority at all, and I don't think they take seriously the idea that there are S-risks downstream of human values (e.g. locking in wild-animal suffering). I'm personally pretty worried about this, not because I have a strong take about the probability of S-risks like this, but because the general vibe is just so apathetic about this kind of thing that I don't trust them to notice and take action if it were a serious problem.
What does "going well" mean?
It seems plausible that many things could be a lot better, like making factory farming obsolete. Does that mean that animals are no longer experiencing extreme suffering? What is our baseline for animal welfare?
I did not realize the stakes were so high...
Wild animal welfare is just a lot more neglected.
Furthermore, while factory farms may go away by default (with the AI transition), the same can not be said about wild animal suffering (unless all wild animals go extinct, which isn't impossible, but... bad vibes).
I think this is a good thing to point out. My main reactions are:
1. I think this work was fairly low-effort and exploratory (how could we get some quick insights using a ton of AI automation), and doesn't have the rigor I think would be needed to draw hard conclusions. For example, the Truth Social data wasn't very high quality.
2. The absence of positive-about-AI content on Bluesky is more stark and statistically significant, and I'm much more confident that a better analysis would turn up that same result.
Survival feels like a very low bar to me.
Survival could mean the permanent perpetuation of extreme suffering, human disempowerment, or any number of losses of our civilization's potential.
Animal welfare is much more neglected than global health (though maybe a bit less tractable).
I really appreciate this post. I also watched this video and was horrified at how no attention at all was paid to wild animal suffering, and I'm really glad there are other people who had the same experience. I think this is a serious moral blindspot in our society.
I'm currently working in technical AI safety, and I have two main thoughts on this:
1) We currently don't have the ability to robustly imbue AI with ANY values, let alone values that include all animals. We need to get a lot farther with solving this technical problem (the alignment problem) before we can meaningfully take any actions which will improve the longterm future for animals.
2) The AI Safety community generally seems mostly on board with animal welfare, but it's not a significant priority at all, and I don't think they take seriously the idea that there are S-risks downstream of human values (e.g. locking in wild-animal suffering). I'm personally pretty worried about this, not because I have a strong take about the probability of S-risks like this, but because the general vibe is just so apathetic about this kind of thing that I don't trust them to notice and take action if it were a serious problem.