I'm quite uncertain about this project, in part because there are so many ways in which a book like this could be written, produced, and shared. However, while I have no prior knowledge of de Haan's work, their MIRI grant and presentation of the proposal passes my gut check for a high-variance creative project like a book.

I appreciate that the author is focused on longtermism and sentientism and that they have an interest in FDT. I wouldn't be too excited about a more generic pitch for such a wide-ranging book, but I think with particular angles like these, it could be more unique and promising. On the other hand, I expect a successful book to usually have a clearer vision and pitch for that uniqueness at the current stage of this project.

I don't have a good sense of what sort of publication is being targeted, and my prior for the success of a self-published book without a significant author platform or exceptionally creative marketing strategy is quite low. I also think two years (even one year) is a long time to write a successful book like this, at least unless there is a lot of original research (e.g., travel, interviews).

I'd be more excited about a smaller grant to develop the idea further and explore alternative strategies like a highly targeted and themed blog, newsletter, forum post sequence, or social media presence. I see my tiny impact certificate offer primarily as a skin-in-the-game upvote for this direction.

Bogard, Caviola, and Lewis are an exceptionally qualified team in terms of academic psychology expertise (e.g., publications in top journals) and their commitment to do the most good.

The general research direction of using psychological theory developed with humans to model LLM and general AI "digital minds" is clearly important. However, I think this is an extremely popular research direction at the moment, and there are many psychologists and behavioral researchers who have already started major research projects and staked out the territory, so I feel skeptical of any new entrants' chances of success—both in terms of generating belief-updating insights and publication.

I think tying in cognitive biases (or other psychology paradigms) to RLHF, DPO, constitutional AI, or other safety-oriented empirical strategies is particularly promising and much less saturated than the more typical research directions. I would encourage the research team to also consider how prompt engineering and in-context methods would affect bias.

It is not clear to me why human subjects research is particularly helpful here. I would need more detail about the methodology, but for now it seems that the three researchers or an informal sample of their colleagues would be a more useful judge of bias and reasoning apparent in LLM output than the samples proposed.

It is also not clear to me why an engineer needs to be hired for this project, which is the exclusive use of funding proposed. Specifically, I don't yet see why a machine learning researcher wouldn't do this in the more typical manner of coauthorship. This could even be a graduate student who is more open to grunt work than late-career researchers. RLHF is messy and expensive, but there are also compute credits available to social benefit research and alternatives to RLHF are increasingly popular.

Chickens are more than 90% of the over 30 billion farmed land-dwelling vertebrate animals and are almost exclusively factory farmed. Humanity should switch to animal-free chicken and egg products, but in the meantime, LIC-style welfare reforms seem among the most effective for reducing chicken suffering. On the other hand, the number of chickens is not so large when weighed against invertebrate and aquatic farmed animals.

LIC is a small, agile organization where I expect donations more cost-effectively address this than at the larger organizations with a wider variety of strategies that are bogged down by various bureaucratic and logistic challenges. On the other hand, overhead can be higher at small organizations if they have less streamlined processes.

Alene has an impressive track record is clearly motivated to address this issue in the most effective ways possible.

Most importantly, supporting cost-effective strategies to reduce factory farm suffering is a promising strategy towards moral circle expansion that will benefit all sentient beings in the long run (e.g., for the more numerous invertebrate populations), particularly by improving the outcomes of advanced AI technologies. Efforts like LIC that build the effective animal advocacy community are particularly promising.