Thanks for writing this Sam! This is a topic I’ve been giving some thought to as I read pro-PLF pro-animal-welfare writers like Robert Yaman (The Optimist’s Barn).
There are two assumptions you make that I think are worth interrogating.
Factory farming cannot be ‘fixed’? Some animal advocates believe that one of the possible end games for animal suffering in factory farming is making welfare so good that animals lives are net positive. I’m unsure if I think this is possible even in principle (it depends on one’s philosophy of wellbeing), but I’m open to it, and if it is, then PLF entrenching an optimised form of factory farming isn’t neccessarily a point against it—in fact it’s exactly what pro-PLF pro-animal-welfare want to happen. We can challenge the possibility of positive welfare factory farming, but I don’t think we can assume it away.
Public advocacy for fixing factory farming in the short-term is counterproductive if our goal is abolishing it in the long-term? I’m far from convinced of this. For example, I think there’s a good case to be made that (a) calling for the abolition of factory farming is so outside the overton window and/or so challenging of most people’s need to see themselves as good-people-that-aren’t-participating-in-a-moral-atrocity that it’s not an effective message for advocates today, (b) calling for reform is a lot more palatable to people, (c) people who are bought into the case for reform today will be more likely to be open to case for abolition tomorrow.
I think you make some strong points in this post though, which I plan to put to pro-PLF folks like Robert Yaman to see what they say. Specifically:
(a) The incentives for industry will remain to maximise profit with welfare as an externality which matters only insofar as it impacts profit due to consumer preferences. Therefore assuming that industry will be willing to trade-off any profit gains for welfare gains is naive, and assuming that using AI to maximise profit will improve welfare (let alone lead to net positive lives) is unjustified.
(b) Managing welfare through opaque blackbox-style optimisation technology, which is developed and deployed too fast for regulators to keep up, is not conducive to holding industry accountable
(c) Using AI towards the PLF end-game for suffering on factory farms instead of the alt-protein end-game for suffering on factory farms seems unwise given one has big downside risks (i.e. increasing total suffering and/or entrenching a food production system that creates net negative lives) and the other doesn’t. We’d need to believe that using AI to advance alt-proteins is far harder to prefer the PLF route, and I haven’t seen a good case for this.
Thanks again!
Thanks for writing this Max! The likelihood that my and other advocates’ work could be made completely irrelevant in the next few years has been nagging at me. Because you invited loose thoughts, I wanted to share my reflections on this topic after reading your write-up:
If AI changes the world massively over the next 5-10 years but there’s no sci-fi-style intelligence explosion:
Many/most of the specific interventions that animal advocates are using successfully today will no longer work in a completely different context.
This means we should ‘exploit’ proven strategies as quickly as possible today (hard with funding as a bottleneck)
This means we shouldn’t ‘explore’ new strategies as much that aren’t robust to a radically transformed world
The most robust strategies for a transformed world (it seems to me) are ones that increase moral consideration that empowered agents (humans and AIs) have for animals, as these will lead agents to make more animal-friendly choices in the world, whatever that world looks like
Unfortunately we’re not very good at this as a movement right now! But more efforts to figure it out, particularly ones that are realistic about human psychology, seem needed to me
If we get an intelligence explosion:
As above, but humans will be making far fewer of the important decisions, and so it becomes far more important to increase moral consideration that AIs specifically have for animals (which means it’s more important for us to target advocacy at the specific people/governments influencing the values of the AIs, and less important to do broad public advocacy)
Either way: AI could make an alt-protein end game for factory farming far more technologically viable. We should be doing what we can to create the most favorable starting conditions for AIs / people-advised-by-AIs to choose the alt-protein path (over the intensification of animal ag path for example). One particularly promising thing we could do here is remove regulatory barriers to alt protein scale-up and commercialisation, because if AI makes it technologically possible but the policy lags behind, this could be reason enough for the AIs / people-advised-by-AIs to decide not to pursue this path.
Keen to hear people’s reactions to this :)