Below is this month's EA Newsletter. I'm experimenting with crossposting it to the Forum so that people can comment on it and give feedback. If this is your first time hearing about the EA Newsletter, you can:
Insects are more complex than you think. Monogamous breeding pairs of termites regularly have 20-year relationships. The smallest mammal brain (a shrew) is only six times the size of the largest insect brain (a solitary wasp). Fruit flies are used in efficacy studies of depression medication — because fruit flies can exhibit depression-like behavior.
Yet scientists have long assumed that insects don’t suffer. As entomologist Meghan Barrett argues on a recent episode of the 80,000 Hours podcast, their reasons for believing this no longer stand up. For example, scientists used to claim that insects didn’t have nociceptors (put simply, pain receptors), but these, and analogous systems, have now been found. It was observed that some insects reacted strangely to grievous injury, ignoring their wounds. Now insects have been observed grooming their own or other insects’ wounds, and even responding to painkillers.
This matters because, in 2020, between 1 and 1.2 trillion insects were farmed, generally without any concern for their welfare. Barrett is too much of a scientist to confidently assert that insects do feel pain. However, she believes that, even if you think insect suffering is unlikely, the sheer number of individuals involved makes insect farming a major issue.
If you are interested in insect welfare, check out Barrett’s related blog post, which provides ways that we can all help to bring awareness to, or work directly on, this problem.
The key is to donate to the right charities. Some are far more effective than others, and some interventions are incredibly cost-effective. A new piece from Our World in Data makes a concise and data-driven case for donating wisely. They argue:
Some health interventions are 1,000x as effective as others (see the graph below). Likewise, the best charities are much more effective at saving lives than the average charity.
Saving a life is relatively inexpensive. GiveWell, a leading charity evaluator, has found four charities that can save a life for around $5,000 — a lot less than the $1 million that the UK government is willing to spend.
If you’re not sure how to pick the best charities, rely on an expert charity evaluator, like GiveWell, or a managed fund. Also, consider sending the article to friends or family who are reluctant to compare charities.
SB 1047 is a California Senate bill that would require companies to develop safety plans for AI models costing over $100 million to train. If companies refuse, they would be liable if the models cause mass casualty events, or over $500 million in damages. The bill has passed the California Senate and now awaits Governor Gavin Newsom's decision to sign or veto.
However, Google, Meta, and OpenAI have been lobbying against the bill, as has Andreessen Horowitz, a venture capital firm heavily invested in AI. Representatives from the tech industry have argued that the bill would stifle innovation among AI startups — even though it only applies to the most expensive training runs — and falsely claimed that it would threaten AI developers with jail.
Crowd-sourced prediction market Metaculus currently puts a 20% chance on the bill being enacted by the first of October this year. Expect more updates in the next couple of weeks.
Alternative protein innovation is in a slump (following a boom). But Lewis Bollard gives some reasons for optimism about the future: governments are investing, retailers are setting ambitious targets, and a recent study shows plant-based chicken nuggets beating chicken nuggets in a taste test.
The Cosmos Fellowship is looking for applicants with “the potential for world-class AI expertise and deep philosophic insight” to work alongside the Human-Centered AI Lab at the University of Oxford (or other host institutions), pursuing independent projects, with access to mentorship. The fellowship pays $75,000 (pro rata) for up to one year of 90-day intervals. Apply before 1 December.
Future Impact Group is offering a part-time, remote-first, 12-week fellowship. Fellows will spend 5-10 hours per week working on policy or philosophy projects on subjects such as suffering risks, coordinating international governance, reducing risks from ideological fanaticism, and more. Apply by 28 September.
Artificial Intelligence Governance & Safety Canada is looking for volunteers to help with content creation, events, translation (French to English) and more. If you’re interested in helping, contact them here.
Conferences and events
Upcoming EA Global Conferences: Boston (1-3 November, apply by 20 October).
Upcoming EAGx Conferences: Bengaluru (19-20 October), and Sydney (22-24 November)
Funding and prizes
The Strategic Animal Funding Circle is offering up to $1 million in funding, to be distributed among promising farmed animal welfare projects. Find out more, and apply for funding before 20 September.
If you are a donor willing to give upwards of $100K to farmed animals per year, you can enquire about joining the funding circle at this email: jozden[at]mobius.life
This is the third in a sequence of posts taken from my recent report: Why Did Environmentalism Become Partisan?
Summary
Rising partisanship did not make environmentalism more popular or politically effective. Instead, it saw flat or falling overall public opinion, fewer major legislative achievements, and fluctuating executive actions.
Public Opinion...
This post presents the executive summary from Giving What We Can’s impact evaluation for 2025. At the end of this post we share links to more information, including the full report and...