The Iliad Intensive

A four-week, full-time course on foundational AI alignment research for mathematicians, physicists, and theoretical computer scientists.

The Iliad Intensive gives you a broad overview of foundational alignment research in areas like deep learning theory, agent foundations, interpretability, and more. It runs in person several times a year, in London and Berkeley.

We are still accepting applicants for our upcoming October and November Intensives, which you can apply to using the link below.

Apply now

Overview

The Iliad Intensive is a full-time, in-person, taught course.

Duration? Four weeks (with occasional three-week iterations).

Schedule? Five days a week, 10am-6pm.

Locations? London (at LISA) and Berkeley.

Cohort size? ~35 participants per cohort.

Cost? No tuition. Participants receive a $5,000 (USD) travel-and-housing allowance, office space at the venue, and lunch and dinner five days a week.

What You’ll Learn

The curriculum is organized into five clusters, each composed of one-day modules. There is more material than fits in a single month, so the exact content varies between iterations.

Alignment: The alignment problem and how to decompose it into subproblems; how frontier labs align models in practice in pretraining, post-training, and deployment; and reward learning theory, including an analysis of various forms of misspecification.

Learning: The mystery of how deep learning works, and components of a theory: singular learning theory, training dynamics, and data attribution.

Abstractions, Representations, and Interpretability: Mechanistic interpretability, computational mechanics and belief-state geometry, natural latents, and an ML engineering introduction.

Agency: Agents with long-term goals and how to model them theoretically: reinforcement learning, idealized agency and AIXI, coherence and expected utility, agent foundations, decision theory, and world models.

Safety Guarantees and their Limits: Debate and its complexity-theoretic foundations, steganography and backdoors, and worst-case interpretability and heuristic arguments.

The full materials and contributors of the April 2026 iteration are available here.

Who Should Apply?

You should apply if you have a strong mathematical background — typically pursuing or having completed a degree in mathematics, physics, or theoretical computer science — and want to work on foundational AI alignment research. We also consider research experience, general competence, and motivation for pursuing the program.

Apply now

What Participants Say

“Genuinely life changing, I could not imagine a better course for understanding AI safety than the Iliad Intensive. Taught by incredible teachers, explaining incredibly tough concepts.”

“To my knowledge, this is a really one-of-a-kind course in goals and content: it filled a niche that I had long felt overlooked, i.e. theoretical upskilling or transition towards AIS research.”

“Crazy pressure cooker for all of the interesting theoretical interp research agendas. New and ambitious, the best place to get a big uplift in both overview and depth of the field.”

“This was the most beneficial learning experience in AI safety I have had so far.”

FAQ

  • The Iliad Intensive is a four-week, full-time, in-person course in technical AI alignment, run by Iliad. It targets people with strong backgrounds in mathematics, physics, or theoretical computer science, and aims to give you both a coherent map of the alignment research landscape and genuine technical depth in topics like singular learning theory, agent foundations, computational mechanics, and debate. Iterations run several times a year, currently in London (at LISA) and Berkeley.

  • The Intensive has similarities to ARENA, but it focuses on foundational AI alignment research rather than alignment research engineering; expect more math and less coding. It is also a taught course, not a research program: the goal is to bring you to the frontier of several research agendas in a month, so that you can then do research in a fellowship (such as the Iliad Fellowship, a research role, or graduate study.

  • There is a dedicated ML engineering day, and several modules include hands-on coding exercises — implementing attention, training sparse autoencoders, and so on. Comfort with Python isn’t strictly necessary, but will make these sessions more productive.

  • Five days a week, 10am–6pm, with lunch and an afternoon break — around 6.5 hours of focused learning per day. Most iterations are four weeks; the August 2026 iteration is a special accelerated three-week version. Plan to treat the program as your full-time occupation for its duration.

  • Yes. We are continuously developing the materials further. 

  • Each day is dedicated to one module and mixes several session types: lectures by module experts, guest lectures and Q&As with external researchers, reading sessions, whole-class and small-group discussions, and math exercise and coding sessions worked alone or in pairs.

  • The materials are created and taught by a team of around twenty internal and external researchers, each with domain expertise in the module they contribute to. Contributors are listed per module in the public course materials.

  • Current cohorts are between 15 and 30 people, but we may adjust cohort size upward in the future if demand is high.

  • Yes, the course is designed for people like you. The Prerequisites module of the public materials lists resources for filling gaps before the program starts.

  • If your strengths lie mainly in engineering, ARENA or similar programs may be a better fit. If in doubt, apply, the application process is simple and quick.

  • No. The requirement is mathematical maturity, not a specific credential.

  • Yes — a single application covers all upcoming programs, and you can indicate which dates and locations work for you.

  • Participants receive a $5,000 (USD) total travel-and-housing allowance per Intensive, office space at the venue, and lunch and dinner five days a week. You arrange and pay for your own travel and accommodation out of the allowance.

  • No. If you cannot attend, the course materials are publicly available for independent study.

  • Feedback on the materials is very welcome — either in the comments of the LessWrong post or by email to feedback@iliad.ac. If you want to contribute to the materials more extensively, please also get in touch.