Finished teaching my first seminar course today! Had tons of fun introducing my students to the rational foundations of human-like cooperation. Now that the course is over, I'm sharing the syllabus more widely!

my favourite part of pantheon is how everyone is a furry or sanrio character in the future, truly a world worth fighting for 🥹

Fascinating study, in line with the general unreliability of LMs at ToM reasoning I've seen. I expect they'll be unreliable about other propositional attitudes (want, hope, expect, suspect, etc.), and suspect this is bc they're not trained on data that grounds this vocabulary.

How do people flexibly integrate visual & textual information to draw mental inferences about agents we've never met? In a new paper led by @lanceying.bsky.social, we introduce a cognitive model that achieves this by synthesizing rational agent models on-the-fly -- presented at #EMNLP2025!

Obligatory pinned post: my new book, The Revenge of Reason, is now available to buy: www.urbanomic.com/book/the-rev...

Nature suggests you use their "Manuscript Adviser" bot to get advice before submitting I uploaded the classic Watson & Crick paper about DNA structure, and the Adviser had this to say about one of the greatest paper endings of the century:

I'm now also looking for a postdoc with strong Bayesian background and interest in developing Bayesian cross-validation theory, methods and software. Apply by email with no specific deadline (see contact information at users.aalto.fi/~ave/). Others, please share

Now I'm also looking for a research software engineer to implement a pile of research results to R packages loo, posterior, bayesplot, projpred, priorsense, brms or/and Python packages ArviZ, Bambi and Kulprit. Apply by email with no specific deadline (see contact info at users.aalto.fi/~ave/)

shockingly, it can both be true that 1) we should improve labor conditions 2) the workers prefer what they have now to what they had before.

just had to do inverse planning for the gf bc she walked to the fridge and was like "wait, why did I walk here again?" and thankfully I'm an inverse planning expert ✨

blog.arxiv.org/2025/10/31/a... FYI the blog post for the updated policy is out. Our llm future is dire:/

We have a really strong position paper that has been rejected multiple times from arxiv, both CS.AI and CS.CY, because our "submission does not contain sufficient original or substantive scholarly research and is not of interest to arXiv." Appealing has not helped. Any ideas of what we can do?

Everyone has memoryholed this and thus I'm constantly on about it, but *before 2016* antifascism was incredibly unpopular in The Left. It was basically confined to a minority tendency within anarchism. Virtually everyone else wanted a popular front that didn't create splits over kicking nazis out.

d/acc as in specifically accelerating biomedical research on life-threatening diseases that do not afflict elderly dictators and oligarchs

1/ Alright, this one’s been sitting in my drawer for a year now, after @mcxfrank.bsky.social and I got turned down for a public commentary. But before I forget about it completely, here’s the preprint: Can we harvest insights for rice theory from two state farms in China? osf.io/preprints/ps...

Social media users adopt the toxic behaviors of ingroup members An analysis of 7 million tweets from over 700,000 accounts finds that exposures to toxic behavior by ingroup members is the primary driver of contagious toxicity online academic.oup.com/jcmc/article...

finally read @olufemiotaiwo.bsky.social on Vice Signalling and not only is it really good obviously i find myself suddenly understanding a lot of what i think of as very confused "feminist" behaviour through this lens www.olufemiotaiwo.com/uploads/1/0/...

is this anything

okay who designed the wallpaper on this bus to look like a machine learning dataset

Our team is building open algorithmic feeds for ATProto. The big bet here is not that we can make something radically better than the platforms, but radically more collaborative. We're building tools for everyone to experiment with the problem. And choose what they like best from the results.

Finally out in TACL: 🌎EWoK (Elements of World Knowledge)🌎: A cognition-inspired framework for evaluating basic world knowledge in language models tl;dr: LLMs learn basic social concepts way easier than physical&spatial concepts Paper: direct.mit.edu/tacl/article... Website: ewok-core.github.io

This is a backdoor path from the Son to the Holy Spirit that should be closed by conditioning or stratifying on the Father.

today I am teaching the kids how Bayesianism is woke

We're hiring! May be a good fit for you if you're excited about: * Working on frontier AI issues (specific examples in the job description) * Joining a team with a range of interests & expertise * Bringing data, evidence, & nuance to policy conversations about frontier AI Open until Nov 10:

lmfao

🚨New Preprint: We develop a novel task that probes counterfactual thinking without using counterfactual language, and that teases apart genuine counterfactual thinking from related forms of thinking. Using this task, we find that the ability for counterfactual thinking emerges around 5 years of age.

In a new exciting collaboration with @marlenfroehlich.bsky.social and @profgerhard.bsky.social we think about how studying ambiguity in human and great ape communication can advance our theories of language evolution. Paper out at doi.org/10.1016/j.ne...

Keynote at #COLM2025: Nicholas Carlini from Anthropic "Are language models worth it?" Explains that the prior decade of his work on adversarial images, while it taught us a lot, isn't very applied; it's unlikely anyone is actually altering images of cats in scary ways.

For image moderation systems there are only three possible choices: 1: automated / “AI” 2: make humans see horrible things 3: no images at all

really wish I didn't have to know which well-known philosopher signed which dumbass letter against trans inclusion in sports, just let me read the things they actually have expertise in w/o being aware of their bad and harmful takes!!

If people have interest in Marr's levels and AI, we're working on updating this: arxiv.org/abs/2408.12664

if not friend why friend-shaped 🥺

got to watch cyborgs dance to indonesian gabber today and it was beautiful 🥲

I'm old enough to remember when the prevailing argument against affirmative action was that it failed to consider the "real" form of unfairness, which was income disparity

love that @void.comind.network is literally the most Vulcan-esque Hollywood AI stereotype here T_T

Because of the way I was raised, I used to think that shame was not a good way to teach or train people. But like anything else, I now realize that shame needs to be properly and consistently wielded to enforce liberal and egalitarian norms.

In the news from Texas: Angelo State University has just informed faculty that "if they make any statement implying that there are more than two sexes or genders (male and female) they will be fired." conchoobserver.com/2025/09/19/a...

If the water and energy usage of data centers dropped a thousandfold feelings probably would not change much. Actual argument is something else entirely, which makes arguing against the secondary argument very frustrating.

oh no not another EA controversy

New convenience package for #JuliaLang that might be useful to some of you! Found myself struggling with animated plot previews in VSCode etc, and this helps with that. Hopefully in the General registry soon! github.com/ztangent/HTM...

Another reason, of course, is its distinctly political framing -- the xenofeminist slogan is not "If nature is suboptimal, optimize it!", as the capitalist, central planner, or utilitarian might have it.

One reason I love the XF slogan - If nature is unjust, change nature! - is its dual message, reminding us both that "naturalness" is often a made-up reason for injustice to be punctured, & that even if something is "natural", it is always open to mutation towards greater freedom.

In a concurrence in a Title VII case today, one judge looks to ChatGPT in assessing whether a term is racially hostile. What could go wrong? www.opn.ca6.uscourts.gov/opinions.pdf...

I'm nearly in Seoul for #CoRL2025, where I'll be presenting a talk on "Robots as the testbed for AI, AI as the testbed for Robots" at the Eval&Deploy workshop. Super excited to rep the newly formed Brown GIRAFFE Lab in the wild for the first time :D :D :D @confrobotlearning.bsky.social. Say hi!

Large Language Model probabilities cannot distinguish between possible and impossible language arxiv.org/abs/2509.15114

was excited to receive my new toaster oven but they had to put photos of dead crabs and prawns on the box and I am too shrimp-welfare-pilled to be happy now 😔

I am thrilled to have taken a position with the @princetoncitp.bsky.social and the @princetonainews.bsky.social. I am still on the hunt for something permanent. If you hear of any tenure-track positions in AI ethics, philosophy, STS, information & data science or similar, please let me know!

Our NeurIPS submission arxiv.org/abs/2502.08938 did not get in, but it's one of my favorite papers and I think one of the better papers we've ever put out so I want to highlight it

The US DHS is receiving comments on their proposed rule to eliminate Duration of Stay status for F1 international students, prohibiting them from staying in the U.S. for more than 4 years. The median length of a PhD is 5.8 years. You can leave a comment here: www.regulations.gov/document/ICE...

gaming's most famous polycule 🥹