Great, thanks Tristan! That's really good to hear, and noted re. the formatting. And yes, we definitely hope that other researchers will build on this and challenge us so that we can continually improve it.

Thanks David! 

• We trialled a few formats, including Notion, and Google Docs was overall the easiest overall for reading and updating, but agree the in-built menu bar is a bit unwieldy. We also considered adding a comparison table but decided not to as distilling the content for each intervention to that extent ended up being unhelpfully reductive. We'll consider other ways to make this easier to navigate though, thanks for flagging!
• Yeah, to keep the document tidy and help us moderate comments then we invite people to directly email me (max.taylor@animalcharityevaluators.org) or Alina Salmen (alina.salmen@animalcharityevaluators.org) with feedback or to request comment access, rather than us enabling comment access automatically.
• Thanks for noting all the cross-over with The Unjournal, that's great! I've added those evaluations to our list of studies to incorporate when we next update this.

Hey Benny, thanks for the thoughts! 

Totally agree on your first point. I guess you could divide positive use cases up into a few different categories:

• In some cases, like alt proteins, there are already people/companies intentionally trying to create less exploitative systems with reasonable chance of success, and AI will help them achieve that.
• In others, like most alternatives to animal testing methods or many forms of Precision Livestock Farming, AI enables methods that are cheaper, more effective, etc., so those methods will probably  end up getting adopted and incidentally helping animals in the process.
• But for many other areas, the positive use cases rely on getting governments, public, industry etc. to care about animals. If we think that we're likely to see transformative AI fairly soon, it probably makes sense to specifically target those kind of MCE efforts at the people who will have most say over AI systems, like governments and AI companies. I've explored that a bit in another post.

On your second point: thanks, that's a good point and I think your suggestion is probably more accurate!

Thanks for this excellent post! The distinction between 'puntable' and 'less puntable' ideas seems like a really helpful way for advocates to think about tactic prioritisation. 

On the point about AI-enabled modelling of wild animal welfare and implications of different interventions: are there any existing promising examples of this? The one example I've come across is the model described in the paper 'Predicting predator–prey interactions in terrestrial endotherms using random forest' but the predictions seem pretty basic and not necessarily any better than non-AI modelling. 

Also, why did you decide that TAI's role in 'infrastructure needs' and 'getting the “academic stamp of approval”' weren't useful to think about?

Nice one, thanks Miranda! Would be really interested to chat about this - I'll DM you :-)

Thanks Vasco, this is a great idea. I'll look into it :-)

That's great to hear! BlueDot has been my main resource for getting to grips with AI. Please feel free to share any ideas that come up as you explore how this applies to your own advocacy :-)

Thanks Tristan! Definitely agree that AGI's effects on animals (like on humans) are currently extremely uncertain – but by being proactive and strategic, we could still greatly increase the probability that those effects will be positive.

The recommendations I suggested seem broadly sensible to me but I'm sure that some are likely to be much more impactful than others, and some major ones are bound to be missing, and each one of them is sufficiently broad that it could cover a whole range of sub-priorities. This is probably an argument for prioritising the first of the principles that you mention, directing the movement toward considering the role of AI in its future, and agreeing on the set of practical, rapid steps that we need to take over the next few years. 

Thanks Simon! Yes, AI for inter-species communication is underway. The main organisations working on this at the moment are Earth Species Project (who just received a $17 million grant) and Project CETI. So far as I can tell, work is still in its early stages and mainly focussed on gathering and cleaning audiovisual data and getting a better sense for different species' portfolio of sounds, rather than actual communication. 

I'm still unsure how good this will be for animals. I wrote a brief post on this for the AI for Animals newsletter if you're interested, but the upshot is that I can see plenty of ways for this technology to be abused (e.g. used for hunting, fishing, exploitation of companion animals for entertainment purposes, co-option by the factory farming industry, etc.). I also think there's a risk that we only use it for communication with a handful of popular species (e.g. dogs, cats, whales, dolphins), and don't consider what this means for other less popular species (like farmed chickens).

The most promising project I've seen so far is the partnership between Project CETI and the More Than Human Life (MOTH) Project at New York University, which is focussed on the ethical implications of interspecies communication. I hope that these kinds of guidelines will end up driving progress on this rather than corporate interests... and that we focus on using AI to understand animals better on their own terms, rather than trying to communicate with them purely for our own curiosity and entertainment.

Thanks! I think your cynical take could be pretty accurate. From what I can tell, the alt protein industry is only making limited use of AI at the moment and no current applications seem like major game-changers. But at least in theory I'd expect increasingly advanced AI to significantly accelerate progress in this area given its potential for speeding up research and development more broadly, so my goal with this research was to try to get a sense for the kinds of specific use cases that might be particularly promising in the future as general AI capabilities improve and as companies/researchers find ways to address the various bottlenecks I mention. There's very limited research on AI and alt proteins and I had to rely a lot on general media coverage, which is obviously pretty limited and skewed, so I'm planning to talk more to experts in the area to get a better sense for this, which I might turn into a follow-up post at some point if it seems helpful.