Artificial General Intelligence: Mathematical Foundations
Cambridge University Press, 2026 (In Progress)
www.agibook.orgSuperintelligence Lab
Advancing mathematical foundations and implementation of safe and reliable ASI
Research
Artificial General Intelligence
Developing mathematical foundations for safe AGI systems with superhuman-level reasoning across domains.
Superhuman Mathematical Reasoning
AI that delivers formal, verifiable proofs for solving IMO problems and university-level mathematics, and collaborates with Mathematicians on open problems.
Self-Improving Systems
Research on autonomously evolving AI systems that continuously improve themselves without losing alignment.
Autonomous Scientific Discovery
Accelerating scientific research through automated hypothesis generation and experimentation.
Continual Meta-Learning
Systems that learn to learn, rapidly adapting to new domains with minimal data while retaining prior knowledge.
AI for Education
Transforming education by personalized learning systems with human oversight.
Human-AI Coevolution
Developing and understanding human-AI interaction and collaboration workflows and their economic intelligence.
Selected Publications
MathEL: Automatic benchmarks with best practices for math competitions
mathel.vercel.appDesigning human oversight in automated scientific discovery
International Conference on Human-Computer Interaction, 2026
idrori.github.io/asdScaling-up autonomous scientific discovery: Lessons from AI studying WASP-4b TTV in TESS data
247th meeting of the American Astronomical Society (AAS), 2026
autottv.vercel.appTransit timing variations of WASP-4b: Decrease in the orbital period
1st Open Conference on AI Agents for Science, 2025
Diverse inference for solving ARC at a human level
1st Open Conference on AI Agents for Science, 2025, (Spotlight)
AI co-scientist for the Habitable Worlds Observatory
Towards the Habitable Worlds Observatory: Visionary Science and Transformational Technology, 2025
AI Innovation Competition
IEEE International Conference on Automatic Face and Gesture Recognition, 2025
Diverse Inference and Verification for Advanced Reasoning
arXiv preprint arXiv:2502.09955, 2025
Human-in-the-Loop AI Reviewing: Feasibility, Opportunities, and Risks
Journal of the Association for Information Systems 25 (1), 98-109, 2024
From Human Days to Machine Seconds: Automatically Answering and Generating Machine Learning Final Exams
ACM Conference on Knowledge Discovery and Data Mining (KDD), 2023
A Neural Network Solves, Explains, and Generates University Math Problems by Program Synthesis and Few-Shot Learning at Human Level
Proceedings of the National Academy of Sciences (PNAS), 2022
The Science of Deep Learning
Teaching
Artificial General Intelligence
Summer 2023, Fall 2023, Spring 2024, Summer 2024, Fall 2024, Spring 2025, Fall 2025, Spring 2026
(๐ 4+/5 formal evals mean, most recent 4.8/5)
Deep Learning
Fall 2017, Spring 2018, Fall 2018, Spring 2019, Summer 2019, Fall 2019, Spring 2020, Summer 2021, Spring 2022, Summer 2022, Fall 2022, Summer 2023, Fall 2023, Spring 2024, Summer 2024, Fall 2024, Spring 2025
(๐ 4+/5 formal evals mean)
Artificial Intelligence
Fall 2022, Spring 2023, Summer 2023, Fall 2023, Spring 2024, Summer 2024, Fall 2024, Spring 2025
(๐ 4/5 formal evals mean)
Machine Learning
Fall 2018, Fall 2020, Spring 2021, Spring 2022, Spring 2023, Fall 2023, Spring 2024
(๐ 4/5 and 5+/7 formal evals mean)
Meta Learning
Fall 2022
(๐ 6/7 formal eval)
Data Science
Spring 2017, Spring 2018, Spring 2019
(NA)
Cybersecurity
Spring 2016, Summer 2016, Spring 2017
(NA)
Data Visualization
Fall 2016, Spring 2017
(๐ highest formal evals)
Optimization
Fall 2017
(NA)
Reinforcement Learning
Fall 2025, Spring 2026
(๐ 4.8/5 formal evals)
Recent Talks
AI for superhuman math
Columbia University, Mathematics Department, Symposium on AI for Mathematical Research
AI for superhuman math and science
Yeshiva University, Mathematics Department
Using AI for solving the IMO and training students
IMO 2025, Sunshine Coast, Australia
Visualizations: P6 answer, ErdลsโSzekeres bound and intersection property
Human oversight of selfโimproving AI researchers
HHMI Janelia Research Campus
Test-time adaptation of foundation models
JPMorgan Chase
AI for solving the IMO at a gold medal level
NVIDIA
AI innovators, startups, and organizations
YPO/Harvard Presidents' Program, hosted by SUNY
Diverse inference and verification for advanced reasoning
Yeshiva University, University of Tennessee Knoxville, University of Birmingham
AI generated classes
MIT Faculty Seminar, Columbia University Senate
Learning to learn courses
MIT-CalTech-UPenn-Stanford NeuroSym Seminar
A neural network solves university math problems at human level
MIT ฯฮฒฯ, Da Vinci Lecture
* Students interested in pursuing a PhD in Computer Science or Mathematics focusing on reliable automated scientific discovery and AI for math at YU with me are invited to contact me directly for a fully funded opportunity.
๐ Yeshiva University
Department of Computer Science & Engineering205 Lexington Avenue
New York, NY 10016
iddo.drori@yu.edu
๐บ๐ธ Stanford University
Department of Management Science & Engineering475 Via Ortega
Stanford, CA 94305
๐ฎ๐ฑ Tel Aviv University
School of Management30 Chaim Levanon Street
Tel Aviv, 6997801, Israel