Ben Stevenson

Animal Welfare Research Assistant @ Rethink Priorities
97 karmaJoined Jul 2022London, UK


I agree: it definitely seems unlikely that plant-based alternatives could substitute for something like live shrimp salad. But note that the majority (~19 trillion) of shrimp killed are A. japonicus (akiami paste shrimp). Relative to other products, paste texture is less complex and the animal origin perhaps less culturally salient; I bet many consumers would substitute shrimp paste for plant-based paste.

(I work w/ Daniela and Elisa but wasn't involved in this report)

Clarifying post, Oliver!

The UK government are using "frontier AI" in their goals for the AI Safety Summit.

And I'll share this comment from OpenAI's Ilya Sutskever:
“My regret is that we’ve chosen to double down on the term AGI,” Sutskever says. “In hindsight it is a confusing term, because it emphasizes generality above all else. GPT-3 is general AI, but yet we don’t really feel comfortable calling it AGI, because we want human-level competence."

Hi Jamie. Extremely interesting post! I'm giving my initial thoughts as an animal welfare researcher who has also participated in animal activism and AI protest.

  • I agree with Tyler that PauseAI, etc., are probably best characterised as moderate in tactics (if not demands)
  • I don't think more radical tactics would be helpful right now, as AI labs and governments are expressing intention to regulate
  • I agree it's important to think about how knowledge flows from/between, e.g., GovAI and PauseAI. Should PauseAI be demanding AI labs meet some of the governance proposals from Schuett, et al (2023)? Does the AI moratorium ask look less realistic when AI public intellectuals like Geoffrey Hinton say it's not doable? If so: does that matter? Or does the radical flank theory apply?
  • There's room in this social change ecology for more actors focused on direct corporate engagement and shareholder activism. I know a couple people are thinking about this
  • Easier said than done: AI corporate campaigners should strive for robust demands that aren't easily Goodharted or 'alignment-washed'. Campaigners should also be wary of alienating potential allies by dismissing any corporate commitments relating to neartermist AI ethics as alignment-washing; corporate advocacy will demand political coalition building 
  • George's work on why companies are motivated to make CSR commitments might be worth reviewing, although he notes methodological limitations
  • Some AI campaigners have experience with animal campaigning, at least in the UK. I think the moral trade idea is interesting!

I've just written a blog post to summarise EA-relevant UK political news from the last ~six weeks.

The post is here: AI summit, semiconductor trade policy, and a green light for alternative proteins (

Early November is the date for the UK’s summit on AI safety, according to leaks yesterday. Offers have been sent out for new AI Civil Service roles. British politics seems increasingly important to the AI safety world.

This is my attempt to justify the ways of Westminster to EA, and EA to Westminster. I’m spotlighting recent headlines on the AI summit, semiconductor trade policy, and alternative proteins.

I'm planning to circulate this around some EAs, but also some people working in the Civil Service, political consulting and journalism. Many might already be familiar with the stories. But I think this might be useful if I can (a) provide insightful UK political context for EAs, or (b) provide an EA perspective to curious adjacents. I'll probably continue this if I think either (a) or (b) is paying off.

(I work at Rethink Priorities, but this is entirely in my personal capacity).

Thanks for writing this; your resource guide is especially helpful!

I agree that we should be thinking about opportunities/harms for animals from current/near-term AI (e.g. machine vision to manage human-wildlife interactions) and from AGI/superintelligence. I’m joining the Slack channel :)

A minor quibble: I agree cultural factors might hold plant-based alternatives back even after reaching taste, texture and price parity, but I am not convinced we currently have alternatives at parity. This Metaculus question resolved ‘annulled’ and while this paper (which found no significant diff in ranking between a plant-based burger and a beef burger) is promising, it would be great to see a replication, more testing for taste and texture specifically, and widespread sales at the same price/cheaper than meat burgers – as well as all those bars cleared again by other plant-based alternatives. I look forward to seeing how AI helps alternative proteins develop.

(in a personal capacity)