David Quarel on AI Safety (Chp. 15)
The reading group returns on Monday the 12th from a holiday break with a visit from David Quarel, who will answer questions on Chapter 15. With a focus on AI safety, this chapter is mostly less technical, so feel free to jump in here even if you missed some (or all) previous chapters. David has…
AIXI with general utility functions: “Value under ignorance in UAI”
This is a linkpost for https://arxiv.org/abs/2512.17086 This updated version of my (Cole Wyeth’s) AGI 2025 paper with Marcus Hutter, “Value under ignorance in universal artificial intelligence,” studies general utility functions for AIXI. Surprisingly, the (hyper)computability properties have connections to imprecise probability theory! AIXI uses a defective Bayesian mixture called a semimeasure, which is often viewed…
Announcing the (Ongoing) Australasian Reading Group on “An Introduction to Universal Artificial Intelligence”
As the first IUAI reading group draws to a close (in a few weeks), we are happy to share that a second round has already started (on Australian/Asian time) thanks to Sigfrido D. Ciletti! Description: Algorithmic Information Theory (Australasia) reading group. Meeting every week, Tuesday at 12pm AEDT (GMT+11:00) [open invitation]. This is also fairly…
Alexander Meulemans & Rajai Nasser on Embedded AIXI
Alexander Meulemans and Rajai Nasser will discuss their work with Google’s Paradigms of Intelligence Team on eMbedded Universal Predictive Intelligence (MUPI): https://www.arxiv.org/abs/2511.22226 Abstract: The standard theory of model-free reinforcement learning assumes that the environment dynamics are stationary and that agents are decoupled from their environment, such that policies are treated as being separate from the world they…
Michele Vannucci on Surprise-Minimizing AIXI
Michele Vannucci is a master’s student at Vrije Universiteit and a regular at the AIXI reading group. This week he will present his work on an interesting AIXI variation that minimizes surprise (in contrast to Orseau’s Knowledge-Seeking Agents) at the regular research meeting (3 pm ET tomorrow, Monday November 30th). Title: Universal Surprise-Minimizing Agents Abstract:We…
Alexis-Walid Ahmed on Applications of AIT to Evolutionary Genetics
Alexis-Walid is a PhD student at Senckenberg who will describe his recent work applying ideas from AIT to evolutionary genetics: Using Hector Zenil’s tools to estimate the Kolmogorov complexity of small strings and matrices, we will see evidence of algorithmicity in genetic network evolution and protein sequences, showing that not only is evolution “aware” of…
Barbara Csima on Algorithmic Randomness
At the research meeting this week (Monday November 3rd at 3 pm ET), Professor Barbara F. Csima will present on “Some Questions of Uniformity in Algorithmic Randomness”:https://www.cambridge.org/core/journals/journal-of-symbolic-logic/article/abs/some-questions-of-uniformity-in-algorithmic-randomness/48F6187C16EA3AB535728FE3942C300E Professor Csima is a mathematician specializing in logic and computability theory at the University of Waterloo. Her talk will be our most technical so far – algorithmic randomness…
Cole Wyeth on Embedded Agency
At the research meeting Monday October 27th, I will present on embedded agency in the AIXI framework, particularly overviewing these papers: https://arxiv.org/abs/2505.17882 https://arxiv.org/abs/2508.16245 and some recent unpublished work: https://www.alignmentforum.org/posts/Go2mQBP4AXRw3iNMk/sleeping-experts-in-the-reflective-solomonoff-prior We will use the usual link: https://uwaterloo.zoom.us/j/7921763961?pwd=TDatET6CBu47o4TxyNn9ccL2Ia8HN4.1 However, the time will be 1 hour earlier than normal, 2 pm ET.
Aram Ebtekar on Algorithmic Thermodynamics
Aram Ebtekar is an algorithmic information theorist and former MATS scholar who has used mathematics to make progress on hard philosophical problems from the arrow of time to AI safety to physics. He will give a talk on algorithmic thermodynamics at the regular research meeting this Monday Oct. 6th at 3 pm ET (note updated…
Q&A with David Quarel on Context Tree Weighting Algorithm
This Monday (October 6th at 12 pm ET), David Quarel will join the reading group on “An Introduction to Universal Artificial Intelligence” (IUAI) to answer questions on the context tree weighting (CTW) algorithm. David Quarel is an author of IUAI who wrote chapter 4 of the book on CTW. He is a PhD student at…
Updated zoom link
Starting with Michael K. Cohen’s upcoming talk at 4 pm ET today, we will be using this new zoom link: https://uwaterloo.zoom.us/j/7921763961?pwd=TDatET6CBu47o4TxyNn9ccL2Ia8HN4.1 This is different from the originally announced link, which had technical issues. The link has been updated in the calendar. Hope to see you there!
Upcoming talk from Michael K. Cohen – and logistical changes!
At our next research meeting, Michael K. Cohen will give a talk on reinforcement learning with KL-regularization to a trusted policy: Reinforcement learning, but don’t do anything I wouldn’t do Cohen is a postdoc at UC Berkeley’s Center for Human Compatible AI. His papers and (highly recommended) blog “Solutions in Theory” argue for the theoretical…
Upcoming talk from Professor Kevin Ellis
Our regular research meeting this Monday (Sept. 8th, 4pm EST = 9pm BST) will begin with a talk on program induction from Professor Kevin Ellis of Cornell University. Computational cognitive scientists1 have invented sophisticated probabilistic programming languages for efficient inference against generative models (e.g. WebPPL, Gen). In his thesis “Algorithms for Learning to Induce Programs,”…
Textbook Reading Group Starts September 1st
Many students and newcomers to UAI are interested in reading Hutter, Quarel, and Catt’s new textbook “An Introduction to Universal Artificial Intelligence.” We have organized a reading group starting Monday September 1st at noon EDT – see the calendar for the most up-to-date information. If you would like to attend the first meeting: Note that…
First Discussion Group Meeting Monday
This is a reminder that our first regular meeting on universal algorithmic intelligence is Monday, August 11th, at 4 pm EDT! (See the event calendar for an authoritative schedule) There is no required reading for this meeting, but you must reach out to an organizer for the meeting link which is not available on the…
Q&A Post
This is the place to ask general questions about logistics and/or UAI. If you want to pose a more detailed technical question, you can also submit it as a blog post!
AIXI and the Anvil Problem
AIXI is to its environment as a player is to a video game. Because AIXI is not computable, it can’t exist as part of a computable universe. This has been raised as a criticism of the theory in this paper and longer blog post on “embedded agency.” Basically, the authors are asking for a theory…
Second Symposium on AIT & ML
(Cross-posted from linkedin announcement) We’re pleased to announce the Second Symposium on Algorithmic Information Theory and Machine Learning, taking place 30–31 July 2025 at Imperial College London, UK. 🧠 Website, Registration, and Schedule: https://sites.google.com/site/boumedienehamzi/second-symposium-on-algorithmic-information-theory-and-machine-learning 🔗 Symposium Info:This symposium builds on the success of the inaugural event held at the Alan Turing Institute in 2022, and…
Universal Algorithmic Intelligence 