iiinova — where your digital self keeps evolving

Reposted by xuan (ɕɥɛn / sh-yen)

Lukas Edman @lukasnlp.bsky.social · 17/09/2026

Ever feel like it's too hard to keep track of what LLMs cannot do as well as humans? We're making your life easier over at: what-llms-can-not-do.github.io We're compiling a list of papers testing the abilities of LLMs against humans. Check it out! And you can help contribute too!

what-llms-can-not-do.github.io

What LLMs Can(not) Do

A living survey of benchmarks that compare large language models with humans.

24516

Reposted by xuan (ɕɥɛn / sh-yen)

Simon Schug @smonsays.bsky.social · 17/09/2026

Human thought is thought to be systematic: Do reasoning models have systematicity of thought? With @brendenlake.bsky.social we study this question in our new preprint arxiv.org/abs/2609.13948 Thread 🧵

14014

Reposted by xuan (ɕɥɛn / sh-yen)

Sev Harootonian @harootonian.bsky.social · 26/08/2026

My main PhD paper is finally out in Nature Human Behaviour! The question: what cognitive strategies do people use when they teach, and can we nudge them toward a better one? Work with my PhD mentors @yaelniv.bsky.social, @markkho.bsky.social and @cocoscilab.bsky.social. 1/5

18318

Reposted by xuan (ɕɥɛn / sh-yen)

Dirk Wulff @dirkwulff.bsky.social · 03/08/2026

🚨 Out now in TICS How can language models help cognitive science? @ruimata.bsky.social & I outline 5 uses: mapping research fields, formalizing theories, cleaning up constructs & measures, predicting behavior across tasks, and capturing environmental variation. 🔗 doi.org/10.1016/j.ti...

03712

Reposted by xuan (ɕɥɛn / sh-yen)

Yoshitomo Matsubara @yoshitomo-matsubara.net · 30/07/2026

Kudos to #ICLR2027 PCs for taking strong actions 👏 I like they keep it open to new people, mitigating a risk of "AI slop" Submission cap should be much lower, say 5 papers per author, unless their contribution is significant and they offer quality review workforce accordingly

1213

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 25/07/2026

I wasn't able to attend #CogSci26 this year, but happy to share that a paper I presented there two years ago is now published in journal-length form 😊 The paper studies how people can rationally infer others' goals in open-ended spaces, and yet do so rapidly and spontaneously.

Topics in Cognitive Science
Original Article
Open Access
Infinite Ends From Finite Samples: Open-Ended Goal Inference as Top-Down Bayesian Filtering of Bottom-Up Proposals
Tan Zhi-Xuan, Gloria Kang, Vikash Mansinghka, Joshua B. Tenenbaum

Open-ended goal inference as top-down Bayesian filtering of bottom-up proposals. We model inference in open-ended goal spaces as the combination of two processes: (a) Bottom-up sampling—you (observer) see your friend (agent) enter the kitchen and come up with several plausible goals (making tea, getting a snack, grabbing drinks, checking the pot, and so on) based on the context cue that your friend is in the kitchen. (b) Top-down filtering—your friend opens the fridge, which would be irrational if they were making tea or checking the pot; these goals are hence filtered out as poor explanations. By repeating these processes as actions unfold, we can model how people draw broadly rational inferences about others' goals even when the space of possibilities is infinite or just too large for explicit enumeration.

1382

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 23/07/2026

relistening to some of Frank Klepacki's music today and only now realizing just how much he was influenced by Rage Against The Machine

130

Reposted by xuan (ɕɥɛn / sh-yen)

Robert Hawkins @rdhawkins.bsky.social · 23/07/2026

excited for the first #cogsci2026 poster session this afternoon! come by and chat with @ymacci.bsky.social @amirzur.bsky.social @clairebergey.bsky.social

1225

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 29/06/2026

"Do not panic or hallucinate like a fragile Large Language Model."

$Question: I'm a baby probabilistic program and Josh Tenenbaum is walking towards me what do I do Answer: Run your generative model backwards immediately to perform inverse planning and infer his goals. As a baby probabilistic program, you represent the "common sense core" of human intelligence. Professor Josh Tenenbaum from MIT studies how minds grow from code. He is likely walking toward you to see if you can pass a Computational Turing Test or prove you possess intuitive physics and psychology. To optimize your survival and impress your creator, execute the following subroutines: 1. Infer His Utility Function (Intuitive Psychology) Do not panic or hallucinate like a fragile Large Language Model. Observe his trajectory, step size, and head tilt. Treat his movements as approximately rational planning under physical constraints. Calculate the probability distribution $P(\text{Goal} \mid \text{Actions})$. If his goal is to test your common sense, prepare to simulate a "game engine in the head".$

1948

Reposted by xuan (ɕɥɛn / sh-yen)

Patient-Led Research Collaborative @patientled.bsky.social · 23/06/2026

We have a new paper out! We asked a simple question about Long COVID “phenotypes": if you take the same patients and the same symptoms but run different clustering algorithms, do you get the same patient subgroups? Short answer: no. 🧵 #LongCovid 1/

12413

Reposted by xuan (ɕɥɛn / sh-yen)

Alejandro Pérez Velilla @apvelilla.bsky.social · 06/06/2026

There, I will be joining the Cooperative Systems and Intelligence Lab under @xuanalogue.bsky.social, where we will work on cool new projects at the intersection of cooperative intelligence/AI alignment and cultural evolution. Onwards to glory!

0121

Reposted by xuan (ɕɥɛn / sh-yen)

Alejandro Pérez Velilla @apvelilla.bsky.social · 06/06/2026

I am happy to announce that I have successfully defended my dissertation --- under the supervision of my mentor @psmaldino.bsky.social --- for the title of Doctor of Philosophy in Cognitive and Information Sciences at the University of California, Merced.

7361

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 03/06/2026

Looking forward to catching up with AI safety/alignment folks and giving a couple of talks at the Center for Human-Compatible AI workshop this week! Also back in Cambridge/Boston after the workshop until early July, if anyone wants to chat :)

0121

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 20/05/2026

Back in Cambridge for a research visit to MIT! Happy to meet and chat with folks who are in town :)

1170

Reposted by xuan (ɕɥɛn / sh-yen)

Cas (Stephen Casper) @scasper.bsky.social · 20/05/2026

🧵🧵🧵 At this point, I think it's only reasonable to fully expect some sort of major and unpleasant AI-caused economic disruption and popular backlash in the next year or two. I'm just not sure which kind.

2143

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 05/05/2026

If you're a late-stage PhD student or post-doc in computer science, and want a free trip to Singapore / NUS, consider applying for this prize: comp.nus.edu.sg/research/nus... Probably helps if you're considering a faculty job at NUS or other universities in Singapore!

comp.nus.edu.sg

Prize - NUS Computing

052

Reposted by xuan (ɕɥɛn / sh-yen)

Alex Turner @turntrout.bsky.social · 28/04/2026

I spent the last 2 months trying to prevent this. If OpenAI offered a fig leaf, Google said "imagine we offered a fig leaf." Google affirms it can't veto usage, commits to modify safety filters at government request, & aspirational language with no legal restrictions. Shameful.

313219

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 31/03/2026

Happy Trans Day of Visibility! 🏳️‍⚧️ This is my first TDoV since moving back to Singapore, and the 12th since I began my transition. It's been a long and sometimes difficult journey, but somehow I've ended up as the first (to my knowledge) publicly out trans professor in the Singaporean academy (cont.)

116315

Reposted by xuan (ɕɥɛn / sh-yen)

Riki Wilchins @rikiwilchins.bsky.social · 18/03/2026

@rikiwilchins.bsky.social 2026 Mar 18 Advocate MIT physics student ordered to vacate campus housing before graduation after his family cut him off financially for coming out as trans raises $85K on GoFundMe in days.

advocate.com

Facing eviction before graduation, MIT student cut off by family raises more than $85K in days

A transgender senior says he lost family support after coming out and was denied financial aid adjustments, endangering his ability to finish school.

0243

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 03/03/2026

frog on the floorrrr where'd he come from nobody knows where he'll goooo

1130

Reposted by xuan (ɕɥɛn / sh-yen)

Mary Elizabeth Sutherland @meharpist.bsky.social · 01/03/2026

Check out this recent @nature.com paper reporting a field experiment on X. It shows X's algorithm boosts conservative content and downranks traditional media—shifting users’ views on key issues. Switching to chronological doesn’t reverse the effect. www.nature.com/articles/s41...

nature.com

The political effects of X’s feed algorithm - Nature

Among users initially on a chronological feed, 7 weeks of exposure to X’s algorithmic feed in 2023 shifted political attitudes and account-following behaviour in a more conservative direction compared...

511853

Reposted by xuan (ɕɥɛn / sh-yen)

Jeremy Berg @jeremymberg.bsky.social · 01/03/2026

NSF Update (Awards through 2/27/26) Directorates to follow 1/10

A line graph showing NSF grant awards made through 2/27/26 for fiscal year 2026 compared with grant awards for fiscal years 2021-2025.

30675445

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 09/02/2026

just had my first encounter with rice weevils that I recall being aware of, and I am simultaneously upset at the fact that: - there were bugs suddenly crawling out of my rice - I have probably already eaten them previously - apparently this is not especially uncommon with rice bought here 😖

210

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 09/02/2026

this prompted me to reread my 1st year college essay on whether proper names w diff. referents are more like indexicals ("I", etc.) vs. homonyms, and I totally forgot I made all these crazy diagrams despite not having taken a philosophy class before truly doomed to be analytic-philosophy-brained 😅

170

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 09/02/2026

Finally bought my first actual copy of "AI: A Modern Approach" (as a reference for the lab) and it's cool to see that assistance games have their own section now! cc: @dhadfieldmenell.bsky.social

2160

Reposted by xuan (ɕɥɛn / sh-yen)

Mar Hicks @histoftech.bsky.social · 06/02/2026

No joke: I got angry hate mail today for writing an obituary of a Black woman scientist—because the person felt she did didn’t deserve the recognition. Which just makes me want to share it again: www.nature.com/articles/d41...

nature.com

Gladys Mae West obituary: mathematician who pioneered GPS technology

She made key contributions to US cold-war science despite facing huge barriers as a Black woman.

13674820319711

Reposted by xuan (ɕɥɛn / sh-yen)

Riccardo Fusaroli @fusaroli.eurosky.social · 04/02/2026

as I'm revising my course materials, I keep stumbling upon cool @mc-stan.org developments. Current favorites: 1. your model has funnels and you exhausted reparametrization ideas: metric = "dense_e" makes your HMC learn about covariance btw parameters. Sloooow, but effective! 1/

23316

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 04/02/2026

Incidentally I think one strong piece of evidence for the theory that (many) emotional expressions and their associated concepts are learned rather than innate is the ever-expanding set of Slack/Discord/etc emoji reactions.

041

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 04/02/2026

I was curious whether people have developed "rational models" of emotional expression, and it looks like there's a very recent paper on the topic! By @yyangteoh.bsky.social and @cendripetalfrce.bsky.social.

Value computations underpin flexible emotion expression

Yi Yang Teoh & Cendri A. Hutcherson

Communications Psychology volume 3, Article number: 169 (2025)

Abstract
Emotion expressions constitute a vital channel for communication, coordination and connection with others, but despite such valuable functions, people sometimes engage in expressive suppression or substitution (expressing emotions they do not genuinely feel). Yet, how exactly do people decide when and what to express? To answer this question, we developed a computational model that casts emotion expressions as value-based communicative decisions. Our model reveals that while people (N = 254) indeed tended to suppress expressions of anger towards others in anticipation of potential social costs as past work theorizes, they also engaged in other nuanced forms of expressive regulation, especially when their reputation was at stake. Most strikingly, people selectively exaggerated/suppressed expressions of happiness when others made more/less equitable choices, seemingly to communicate stronger normative preferences for fairness than they privately held. Together, these findings yield insights into how people regulate their emotion expressions, providing a mechanistic and unified account of the different expressive behaviors people flexibly engage in to navigate their complex social interactions with others.

160

Reposted by xuan (ɕɥɛn / sh-yen)

Yang Teoh @yyangteoh.bsky.social · 24/11/2025

Happy to finally see this paper (w/ @cendripetalfrce.bsky.social) out now @commspsychol.nature.com We built a computational framework to show people flexibly balance communicative benefits against social costs when deciding whether and how to express their emotions. www.nature.com/articles/s44...

nature.com

Value computations underpin flexible emotion expression - Communications Psychology

People do not always express the emotions they feel truthfully. Computational modelling reveals that people flexibly regulate their emotion expressions by balancing their value as a communicative sign...

2124

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 04/02/2026

One of the earlier feminist lessons I had to learn in life was how to cut off friendships with sexual abusers (+ their friend groups). I unfortunately had to learn this in high school, but with all the Epstein stuff coming out, I'm reminded that most haven't learned this at all.

050

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 29/01/2026

in which my students explain the context-sensitivity of social roles and norms 😅

1151

Reposted by xuan (ɕɥɛn / sh-yen)

Hadas Weiss @hadas.bsky.social · 27/01/2026

a phd

1486799

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 27/01/2026

other people: reading the Innovator's Dilemma bc they want to run a successful start-up me rn: reading the Innovator's Dilemma bc I'm a historical materialist 🌹

071

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 26/01/2026

I'd like to say something about ICE, but what if they review my social media history the next time I apply for an ESTA, decide my wokeness will destroy America, and shoot me 10 times dead when I next cross the border? Can't really blame them 🤷🏻‍♀️

080

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 24/01/2026

watched Ed Feigenbaum talk at #AAAI2026 and apparently the way he got into AI research was kinda like mine --- got bored as an elec. eng. freshman, took a class on "Ideas & Social Change", which led to studying with Herbert Simon on Math. Models in the Social Sciences, and hence, AI!

061

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 12/01/2026

Just walked past a NUS student dance group doing warmups to a ballroom / vogue track, Ha Dub and all, and frankly I'm confused how the gayest possible dance music has become this globally mainstream.

140

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 06/01/2026

I regret to report that the AI ads have spread to HK airport.

120

Reposted by xuan (ɕɥɛn / sh-yen)

Jenna Norton @jenna-m-norton.bsky.social · 06/01/2026

The district court injunction that prevented NIH from capping indirects at 15% was upheld today on appeal! www.reuters.com/world/trump-...

reuters.com

Trump administration cannot slash NIH research funding, court rules

District Judge Angel Kelley last year blocked the cuts, and on Monday the appeals court agreed.

8505170

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 06/01/2026

bye again america please invade less countries next time if/when I visit

0300

Reposted by xuan (ɕɥɛn / sh-yen)

Ted Underwood @tedunderwood.com · 05/01/2026

This is an important negative reality. It's also sustained by industry incentives—and the thing most likely to break it is a big company or two deciding it's bad business to network on a competitor's site. I am a fan of banning mean bsky account FlameTroll420, but that's not the tipping point.

1345

Reposted by xuan (ɕɥɛn / sh-yen)

John Herrman @jwherrman.bsky.social · 05/01/2026

Wrote about an obvious and yet profoundly underappreciated aspect of the AI boom: its total narrative capture by Elon Musk's X nymag.com/intelligence...

The reality of the AI industry is far larger than a subculture on X with hundreds of billions of dollars flowing into physical infrastructure, reshaping jobs, and deploying into products used by billions of people. But that makes X's capture of the AI conversation all the more notable. Much as Twitter long ago captured, amplified, and distorted elite conversation in media, in sports, in parts of finance, in left politics, and more recently in right politics, the world's understanding of what's going on in Silicon Valley right now - the hype and the doom and the bubble and the progress - is first processed through the strange culture and incentives of X, which is now owned by estranged OpenAI co-founder, Google antagonist, and xAI founder Elon Musk. It has been merged into xAI, where the in-house chatbot occasionally confuses itself with Hitler. When the history of this period in tech is eventually written, it will be composed of, among other things, a whole lot of threads and replies from blue checks on X.

Then, of course, there's the Musk factor. The fact of his ownership creates obvious risks for all of his competitors, who rely on his platform to supply the vibes they need to help keep their companies, and the AI boom in general, going. It's worth remembering that, back before he decided to purchase it and let it drive him nuts, Twitter was, to Musk, a platform with clear functional utility, a social network that helped make him wildly famous and use that fame to convince thousands of retail traders to buy into Tesla, the company that made him the richest man in the world.
Even if we assume, against all reason, that Musk won't leverage this unusual sort of power over all the competitors gathered on his platform, X's influence over them and how they're seen by the rest of the world is undeniable. Along with "AI Twitter," some of the other communities thriving on Musk era X include MAGA guys and crypto guys, meaning that the background vibe of the primary gathering place for the industry that's trying to remake the world is quite a bit more conservative and mercenary than it otherwise might have been, and while the primary factor in the public's deep skepticism of where Al is going is clearly fear about labor, association with polarizing Muskian politics probably doesn't help assuage ambient fears about job loss and surveillance. (One important consequence of having the AI elite gathered on X is that their media diets are now substantially made up of one another's posts, effusive and dismissive responses to their oren posts, and standard "For You" page Musk-era algo-slop.
It's hard to overstate how thoroughly their information diets are influenced by their timelines, and you can get a pretty good idea of what those look like by jacking into
AI Twitter yourself.)

736095

xuan (ɕɥɛn / sh-yen) @xuanalogue.bsky.social · 05/01/2026

love listening to a random Vivaldi mandolin concerto for the "first" time, finding the motif strangely familiar, then realizing I actually first heard it as a sample of a Yo-Yo Ma cello cover in a 90s hip-hop album

150

Reposted by xuan (ɕɥɛn / sh-yen)

Natacha @natacha.bsky.social · 05/01/2026

Guardian reports how misgendering can affect mental health when the victim is cis. Doesn't seem so bothered when it happens to trans people. www.theguardian.com/world/2026/j...

Brigitte Macron’s daughter Tiphaine Auzière, 41, a lawyer, had told the court that false claims the French first lady was born a man had damaged her mother’s quality of life, leaving her worrying every day about the clothes she wore and how she stood.

Auzière, who is one of three children from Brigitte Macron’s first marriage, said her mother had been affected by social media posts that had caused a “deterioration of her health” and a “deterioration of her quality of life”.

4541144

Reposted by xuan (ɕɥɛn / sh-yen)

Colin @colin-fraser.net · 03/01/2026

Again I find it baffling to compare the coverage to when Gemini generated implausibly many Black hockey players etc. That was apparently a societal emergency. But the mass harassment campaign taking place in public barely warrants a mention.

051967

Reposted by xuan (ɕɥɛn / sh-yen)

Colin @colin-fraser.net · 03/01/2026

I know this is yesterday’s news and feels relatively unimportant now but just want to note that this is ongoing and X and xAI continue to do absolutely nothing about it despite being able to end it by pressing a single button.

655295