Advanced Topics in Rational Intelligence: Prediction, Causation, Decision, Incentive, and Regulation All at Once

 

Overview

This seminar explores research themes at the core of the Rational Intelligence (RI) Lab (https://ri-lab.org/), spanning prediction, causation, and decision-making, as well as incentive-aware learning and the regulation of algorithmic systems, including public, open-sourced, and proprietary systems owned by big companies.

At its heart lies a central question: How can these techniques help us build truly intelligent systems that generalise reliably in heterogeneous and non-stationary environments under distribution shifts, diverse preferences, and potentially misaligned incentives?

The topics of interest include, but are not limited to:

1. Prediction under heterogeneous and dynamic environments: Distribution shift and domain generalization; performative prediction; feedback loops; in-context learning; meta-learning; robustness to non-stationarity; compositional generalization; learning under partial observability.
2. Causal and counterfactual inference: Causal machine learning; structural causal models; instrumental variables; proxy variables; partial identification; sensitivity analysis; transportability and external validity; causal discovery under uncertainty.
3. Decision-making under uncertainty: Model-assisted and human-AI decision making; robust and distributionally robust optimization; imprecise probabilities; ambiguity-aware learning; prediction and decision markets; Bayesian and non-Bayesian updating; epistemic vs. aleatoric uncertainty.
4. Regulation and governance of AI systems: Mechanism design for AI; incentive-aware learning; AI alignment; algorithmic auditing; explainability and accountability; threshold-based regulation; strategic behavior and gaming; multi-agent learning and equilibrium analysis.

Format

In this seminar, students will develop core research skills by critically reading and analyzing research papers on selected topics, preparing and delivering presentations, and engaging in in-depth discussions with fellow participants and RI lab members.

The students will be assigned one topic out of the above areas. Based on their preference and academic background, they will then pick (or are assigned) 1-2 papers to study in detail, including the related literature. After studying the paper(s), they must give a presentation about the paper to the rest of the class. Finally, the students submit the final report and presentation.

During the presentation, the students will be assigned one of these two roles:

• Presenter: The presenter’s role is to communicate the paper clearly and coherently to the class, demonstrating a solid understanding of its technical content and responding effectively to questions from the audience.
• Discussant(s): The discussant’s role is to lead the post-presentation discussion of the assigned paper, offering a critical assessment of its technical novelty, advantages over existing work, and limitations.

While presenter assignments will be announced in advance, discussants for each presentation will be selected on the spot.

Schedule

The seminar presentations will take place weekly on Thursdays from 14:00-16:00 from the first day of presentations. Participation in all sessions is mandatory.

• The kick-off meeting (April 16, 2026).
• The first presentation (TBA)
• The deadline for the final report and presentation (TBA)

Date	Time	Title	Presenter	Discussant(s)
Friday, May 15	16:00-18:00	Collaborative Causal Inference with Fair Incentives	Maike Kalms	TBA
Thursday, May 21	14:00-16:00	A Review of Generalizability and Transportability	Haniya Fatima	TBA
Thursday, May 28	14:00-16:00	On Counterfactual Metrics for Social Welfare: Incentives, Ranking, and Information Asymmetry	Abdul Ahad	TBA
Thursday, June 11	14:00-16:00	Large Language Monkeys: Scaling Inference Compute with Repeated Sampling	Stefaniia Ovcharenko	TBA
Thursday, June 18	14:00-16:00	Eliciting Human Preferences With Language Models	Zirui Wang	TBA
Thursday, July 16	14:00-16:00	E-Scores for (In)Correctness Assessment of Generative Model Outputs	Hlib Kriukov	TBA
Thursday, July 23	14:00-16:00	CausalPFN: Amortized Causal Effect Estimation via In-Context Learning	Shilpa Suresh	TBA
Thursday, August 6	14:00-16:00	Concrete Problems in AI Safety	Timm Groß-Selbeck	TBA
Thursday, August 13	14:00-16:00	Possibilistic Inferential Models: A Review	János Csicsely	TBA
Thursday, August 20	14:00-16:00	Uncertainty Measures: The Big Picture	Saba Noorghasemi	TBA

 

If you would like to swap your presentation date, please contact the person you wish to exchange slots with directly. Please also inform us so that we can update the schedule accordingly.

Location

• CISPA C0 - 0.07 Meeting Room on the university campus, except on 07/05, 02/07, and 20/08. The location for 07/05, 02/07, and 20/08 will be announced in due time.
• Under exceptional circumstances that prevent you from participating in person, you may join via the following Zoom link. 
  • Zoom link: https://cispa-de.zoom-x.de/j/61914314679?pwd=15b2Qqh6hb5PUkL9rbYG63bS8EazEh.1
  • Passcode: !y0c?A

Papers

1. Collaborative Causal Inference with Fair Incentives (https://proceedings.mlr.press/v202/qiao23a.html)
2. A Review of Generalizability and Transportability (https://www.annualreviews.org/content/journals/10.1146/annurev-statistics-042522-103837)
3. General Transportability – Synthesizing Observations and Experiments from Heterogeneous Domains (https://ojs.aaai.org/index.php/AAAI/article/view/6582)
4. CausalPFN: Amortized Causal Effect Estimation via In-Context Learning (https://openreview.net/pdf?id=RblaNJGx8C)
5. Foundation Models for Causal Inference via Prior-Data Fitted Networks (https://openreview.net/pdf?id=d2L1ndOKjq)
6. Leaderboard Incentives: Model Rankings under Strategic Post-Training (https://arxiv.org/abs/2603.08371)
7. E-Scores for (In)Correctness Assessment of Generative Model Outputs (https://arxiv.org/abs/2510.25770)
8. On Counterfactual Metrics for Social Welfare: Incentives, Ranking, and Information Asymmetry (https://proceedings.mlr.press/v238/wang24b.html)
9. Performative Prediction on Games and Mechanism Design (https://arxiv.org/abs/2408.05146)
10. Learning Optimal Contracts: How to Exploit Small Action Spaces (http://arxiv.org/abs/2309.09801)
11. Mechanisms that Incentivize Data Sharing in Federated Learning (http://arxiv.org/abs/2207.04557)
12. Scaling Can Lead to Compositional Generalization (http://arxiv.org/abs/2507.07207)
13. Game-Theoretic Statistics and Safe Anytime-valid Inference (http://arxiv.org/abs/2210.01948)
14. Fiducial Inference Viewed through a Possibility-Theoretic Inferential Model Lens (https://proceedings.mlr.press/v215/martin23a.html)
15. Possibilistic Inferential Models: A Review (http://arxiv.org/abs/2507.09007)
16. Proof-of-Learning: Definitions and Practice (https://ieeexplore.ieee.org/document/9519402/)
17. Uncertainty Measures: The Big Picture (http://arxiv.org/abs/2104.06839)
18. Valid Inferential Models for Prediction in Supervised Learning Problems (http://arxiv.org/abs/2112.10234)
19. Robust Predictions in Games With Incomplete Information (http://doi.wiley.com/10.3982/ECTA11105)
20. Eliciting Human Preferences With Language Models (https://arxiv.org/abs/2310.11589)
21. Large Language Monkeys: Scaling Inference Compute with Repeated Sampling (https://arxiv.org/abs/2407.21787)
22. Concrete Problems in AI Safety (https://arxiv.org/abs/1606.06565)