Anthropic projects 80% of its revenue this year will come from selling AI models to business customers via API, contributing to its projected cost efficiency. Full story here: https://thein.fo/43kkoma

Anybody else increasingly mystified?

Thank you to Paul Hünermund @p-hunermund.com , Jermain Kaminski @jermainkaminski.bsky.social, Carla Schmitt, and Beyers Louw for organizing the #CDSM25 event again this year. Well facilitated, great speakers and I loved the 80 minute poster talk sprint! I learned a lot.

So I met this guy who’s meditated 4000 conversations between Israelis and Palestinians since October 7th. And he’s building an AI powered platform for difficult conversations. Here’s what he’s learned. www.betterconflictbulletin.org/p/what-you-c...

Finally! A positive incentive to tell the truth! Lee, J. Y., Shin, J., & Yu, J. (2025). Communicating attribute importance under competition. Marketing Science.

Coming up: www.causalscience.org #CDSM25

Coming up in a few hours (16h20 CET): Generating Impact with Causal ML: Applications, Strengths, and New Opportunities w/ Stefan Feuerriegel (LMU) www.causalscience.org #CDSM25

Coming up! www.causalscience.org #CDSM2025

"Check yourself"

King, J., et al (2025, October). User Privacy and Large Language Models: An Analysis of Frontier Developers’ Privacy Policies. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (Vol. 8, No. 2, pp. 1465-1477).

" If you want to show that research should be done, why not do it and add a discussion to your paper? A demonstration, no matter how small, can be done through an experiment or through some mathematical and theoretical development."

Tomorrow's newsletter: Looped Language Models, less superhuman, privacy, attribute importance, political connections, musky drama, economic lit search

Love this paper. Bak-Coleman, J., O'Connor, C., Bergstrom, C., & West, J. (2025). The Risks of Industry Influence in Tech Research. arXiv preprint arXiv:2510.19894.

Shares of Duolingo vs. the S&P. See if you can spot the day Apple announced real-time language translation. 👀 $DUOL

Bak-Coleman, J., O'Connor, C., Bergstrom, C., & West, J. (2025). The Risks of Industry Influence in Tech Research. arXiv preprint arXiv:2510.19894.

Hard interviews signal greatness to great people. Ash, Shukla, Sockin (2025) Interviews

AgentFold treats memory as clay, not tape. It folds, condenses, and abstracts — a glimpse of long-horizon cognition.

In tomorrow's newsletter: Inside the Fold. How agents sculpt memory, maps reveal bias, and industries shape their own evidence.

"Frightening away demons is the reason for the season!"

Another cool one! Hewitt, L., Ashokkumar, A., Ghezae, I., & Willer, R. (2024). Predicting results of social science experiments using large language models. Preprint.

More from Peng et al. So cool.

Peng, T., Gui, G., Merlau, D. J., Fan, G. J., Sliman, M. B., Brucks, M., ... & Toubia, O. (2025). A Mega-Study of Digital Twins Reveals Strengths, Weaknesses and Opportunities for Further Improvement. arXiv preprint arXiv:2509.19088.

Neat.

Signal. Brand, J., Israeli, A., & Ngwe, D. (2023). Using LLMs for Market Research (Harvard Business School Marketing Unit Working Paper No. 23-062). Social Science Research Network. doi. org/10.2139/ssrn, 4395751.

A better way to bandit. Old trick. New arms. Weaver, I. N., Kumar, V., & Jain, L. (2025). Nonparametric Pricing Bandits Leveraging Informational Externalities to Learn the Demand Curve. Marketing Science.

Wild. Pio-Lopez, L., Hartl, B., & Levin, M. (2025). Aging as a loss of goal-directedness: an evolutionary simulation and analysis unifying regeneration with anatomical rejuvenation.

Lies. Lies. Popular lies. El, B., & Zou, J. (2025). Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audiences. arXiv preprint arXiv:2510.06105.

We're going hard on this idea of Moloch's Bargain for AI. El, B., & Zou, J. (2025). Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audiences. arXiv preprint arXiv:2510.06105.

Anti-scheming research stress-tests AI alignment. Even with interventions, covert behaviors persist (from 8.7% → 0.3%). Alignment isn’t just preventing failure: it’s guarding against hidden goals.

GOFFMAN! Cheng, M., Yu, S., Lee, C., Khadpe, P., Ibrahim, L., & Jurafsky, D. (2025). Social sycophancy: A broader understanding of llm sycophancy. arXiv preprint arXiv:2505.13995.

TeachLM asks: what if LLMs acted like tutors, not assistants? Instead of spoon-feeding answers, tutoring requires friction: withholding, probing, adapting to a learner’s state of mind.

How small can a language model be and still tell a coherent story? Just 2.5M params can spin simple, fluent narratives on par with models 10–100x larger. Eldan, R., & Li, Y. (2023). Tinystories: How small can language models be and still speak coherent english?. arXiv preprint arXiv:2305.07759.

We're covering TinyStories in tomorrow's newsletter Eldan, R., & Li, Y. (2023). Tinystories: How small can language models be and still speak coherent english?. arXiv preprint arXiv:2305.07759.

Happy Thanksgiving to all who celebrate.

Meincke, L., Shapiro, D., Duckworth, A., Mollick, E. R., Mollick, L., & Cialdini, R. (2025). Call Me A Jerk: Persuading AI to Comply with Objectionable Requests. Incredible that this works as well as it does!

Seems odd that they'd go closed... curious.

Hummel, J. E., & Heaton, R. F. (2025). From Basic Affordances to Symbolic Thought: A Computational Phylogenesis of Biological Intelligence. arXiv preprint arXiv:2508.15082.

Taleb's point about local governance is well expressed. Spicy. Spicy post.

In tomorrow's newsletter. This Recht guy has a lot of good ideas. Almost James March-y in the clarity. I'm enjoying him a lot. Maybe too much.

What will be the main bottlenecks for growth?

Great paper. The power concentration question is a fantastic one. Those with it tend to justify it on the basis their judgement on allocating resources is superior to others. And...well, I'm wondering what you think of their self-assessment?

Seriously appreciating the approach the QWEN team took to safety. Solid.

In this week's newsletter: Do markets believe in transformative AI, research agenda for TAI, behaviors, the sims, Qwen3Guard

How might AI shift political power dynamics?

Time for another re-org. What does the lit say about that? Joseph, J., & Sengul, M. (2025). Organization design: Current insights and future research directions. Journal of Management, 51(1), 249-308.

Replacing humans with AI is harder than it seems. "The algorithm will see you now" worksinprogress.co/issue/the-al...

Is there broad awareness of data quality issues that holding back AI integration initiatives? Or is there just broad based denial? Anomalous, right?

PSI is in the right direction. Kotar, K., Lee, W., Venkatesh, R., Chen, H., Bear, D., Watrous, J., ... & Yamins, D. (2025). World Modeling with Probabilistic Structure Integration. arXiv preprint arXiv:2509.09737.

Anything that resembles backcasting is going to be at least good. This, however, is great. arxiv.org/abs/2509.06160

LLM's are just like us! Abdelsalam, Y., Peitek, N., Maurer, A. M., Toneva, M., & Apel, S. (2025). How do Humans and LLMs Process Confusing Code?. arXiv preprint arXiv:2508.18547.