Enjoyed talking on #AIwisdom at Jnana Prabhodini and IISER in Pune, India.

free access for the next 50 days here:

doi.org/10.1016/bs.a...

#CulturalChange #SocialPsychology #TimeSeries #CulturalEvolution #Psychohistory

9/ This chapter is dedicated to the memory of Mark Zanna, long-time AESP editor, my friend and role model at Waterloo. I think of him whenever I think of this outlet.

Baby steps toward Psychohistory.

8/ Looking ahead:
• Forecasting & retrodiction tournaments
• Culture-ecology feedback loops
• Historical Large Language Models—AI trained on historical texts to simulate the psychology of the dead
Sci-fi? Maybe. But it was sci-fi when two boys read Asimov’s Foundation on different continents.

7/ Perhaps most humbling: social scientists are bad at predicting cultural change. In our forecasting tournaments during COVID, experts performed no better than laypeople — and sometimes worse than a model that just guesses ‘same as last year.’

6/ → Pop songs are getting simpler. And simpler songs chart higher when there’s more music competing for attention.

→ The collective attention span of academia is shrinking — keyword half-lives dropped 50% in two decades. We’re not immune to the same forces we study.

5/ → Gender equality gains follow declines in infectious disease, with a 15-25 year lag.
→ Population density predicts falling fertility in 174 countries over 7 decades. Holds after controlling for GDP, education, contraception, religion, etc.

4/ → Individualism is rising worldwide — and it tracks rising wealth, declining disease, and urbanization. Not just a Western story. Replicated across 50+ countries.

→ But it can reverse. In Germany, the long trend toward individualism temporarily flipped during the Nazi era.

3/ So we built an ‘ecological time-series approach.’ Using census data, text corpora, and surveys spanning decades–centuries, we link cultural shifts to changes in ecology: resources, disease, density.

Key findings:

2/ I grew up during the collapse of the Soviet Union. Values and norms rearranged. That left a mark.

It also left a question: why does soc psychology—a field obsessed with change—almost never study change over time? Most evidence = single snapshots. That’s like understanding a movie from polaroids.

1/ NEW in Advances in Experimental Social Psychology — our first review of a decade+ of research on understanding & predicting cultural change, with Michael Varnum.
This one is personal. A thread on what we found, what surprised us, and how two kids reading Asimov ended up here. 🧵

Open questions I don't have answers to: Can AI review policies be enforced when denial is costless? Time for signed reviews? Do we need norms for AI-assisted vs AI-delegated reviewing? Or did we build a system so reliant on unpaid labor that this was inevitable? #AcademicChatter #OpenScience

The real problem: AI writes papers. Another AI reviews them. An editor uses AI to summarize reviews. The author uses AI to read the decision. A BS circle of post-reasoning academic life. Are we still in the loop or are we getting loopy? #PeerReview #Reviewer2

Flagged it to the journal. Response: 'no direct evidence of a violation, AI for formatting is permitted.' I get it; as a senior editor myself I know we're all overwhelmed. Policies against AI-generated reviews are functionally unenforceable. And denial is costless.

The signs: title-cased section headers. Unicode characters (≈) that are ChatGPT defaults but don't exist on a standard keyboard. A polished structure matching 'write a negative review of this manuscript.' Fluent but generic critique that could apply to literally any empirical paper.

Recently got 5 excellent reviews on a paper. Super thankful — four reviewers clearly put real thought into strengthening the work. The problem: Reviewer 2 (of course!) was, by every available indicator, substantially generated by a large language model. 🧵

Imagining and building wise machines: the centrality of AI metacognition

By Samuel Johnson @amirhkarimi.bsky.social @yoshuabengio.bsky.social @tobigerstenberg.bsky.social @sydneylevine.bsky.social @melaniemitchell.bsky.social @iyadrahwan.bsky.social @igi.bsky.social and more

tinyurl.com/36yc6ant

What is wisdom, and can it be taught? Scientists are trying to name the qualities that make someone wise and figure out how to cultivate them

From Knowable Magazine: What is wisdom, and can it be taught? | Knowable Magazine knowablemagazine.org/content/arti...

Psychological Inquiry Volume 36, Issue 4 of Psychological Inquiry

www.tandfonline.com/toc/hpli20/c...

Is the Theory of Mind paradigm simply hateful, a persistent myth, or does it require meta-analytic convergence? Decide for yourself! 📖 The target article and at least one of the rebuttals are OPEN ACCESS!

The rebuttals are FIRE! 🔥 We have serious, rigorous pushback and commentary from major voices in the field. Hear what Baron-Cohen, @utafrith.bsky.social , @gernsbacherlab.bsky.social, Joe Gough, Patrick Dwyer & others have to say about the debate!

OUT NOW: The highly controversial issue of Psychological Inquiry (Vol. 36, Issue 4) is live!
Are we rethinking everything we know about autism? In our newest target article, @travislacroix.bsky.social argues that the "Theory of Mind" paradigm in #autism research is actually pseudoscience. 🧠 🧵👇

@tobigerstenberg.bsky.social @iyadrahwan.bsky.social @ktlr.bsky.social

link to the paper here: doi.org/10.1016/j.ti...

3/3 As the figure shows, applying a wisdom lens to AI helps regulate intractable problems, making systems more robust, explainable, cooperative, and safe. Let's move beyond static alignment! 👇

A roadmap for evaluating moral competence in large language models - Nature This Perspective offers a roadmap for tackling the challenges of the facsimile problem, moral multidimensionality and moral pluralism in large language models.

2/3 The recent @nature.com roadmap for AI moral competence showed we must evaluate how AI reasons morally. www.nature.com/articles/s41... We argue the missing piece to achieve this is perspectival metacognition: teaching AI intellectual humility, context adaptation, & viewpoint balancing.

1/3 AI is getting smarter, but is it getting wiser? 🤔 Thrilled to share our new @cp-trendscognsci.bsky.social paper on building wise machines, co-authored with Sam Johnson, @amirhkarimi.bsky.social, @yoshuabengio.bsky.social , @sydneylevine.bsky.social , @melaniemitchell.bsky.social, & more!

Full paper draft under review in Annual Review. Navigating the Unknown: The Sociocultural and Metacognitive Architecture of Wisdom @SSRN papers.ssrn.com/sol3/papers....

12/12 This synthesis moves us from wisdom as a mystical and rare ideal to a tangible, highly practical construct which couldn't be more relevant in the uncertain world of today.

11/12 Wisdom is a dynamic system. It includes the Intergenerational Database (Culture) accessed by a High-Fidelity Search Engine (Metacognition).

To foster it, we must get a handle on both cultural evolutionary and cogsci angles to the issue.