January 2025 Newsletter 💫🎱🌳 Active Inference Institute
Updates from the Institute at the start of 2025
Happy New Year! Welcome to the January 2025 newsletter from the Active Inference Institute.
Here is the entry point for 2025, to our main living document. There are many ways to get involved with ongoing activities, and we are excited to see how the Institute and Ecosystem develop this year.
Read on for many updates, and all are welcome to send in a Measurement for possible inclusion in future newsletter.
Institute scale
As we begin the year, we welcome familiar and new people into the following positions. Thank you to all those who have stepped up into these volunteer service roles:
The Scientific Advisory Board (SAB) for 2025 has been assembled. With 43 members, this is our 4th and largest cohort yet! We are excited to see how SAB members provide scientific insight and guidance across different areas of research, education, and service at, and with, the Institute.
The Board of Directors for 2025 have been selected: John Clippinger, Bleu Knight, Alex Vyatkin, Daniel Friedman, Mike Smith, and Vladimir Baulin.
The Officers for 2025 have been elected: Daniel Friedman (President & Treasurer) and Alexandra Mikhailova (Vice-President and Secretary)
We are continuing early collaborative development on a possible Active Inference Summer School, to be held later in 2025. Please reach out if you are interested in being a physical host, facilitator, sponsor, or other partner for this effort.
We have created an Affordances page, listing targeted opportunities for contribution.
The Theoretical Neurobiology (TNB) Group is now hosted at the Institute. Information on how to join these weekly meetings can be found at the project page. We are honored and excited to steward the continuation of this historic forum.
EduActive
The Textbook Group reports: We completed Cohort 6 and half of Cohort 7 during 2024. Cohort 7 will continue, and Cohort 8 will begin in February. Registration is open for learners of all backgrounds and familiarity with Active Inference — this is a great opportunity to learn fundamentals, stay updated with applications, and connect with other learners.
The Audio-Visual Production project reports: We made 137 videos from 2024, reviewed in 2024 ReviewStream. We are currently open to interest for people who are interested to contribute to recorded & livestreamed materials in 2025 (watch this short video with more information).
The Active Inference Journal reports: This year we seek to continue open source methods developments on Journal-Utilities, and publications of transcripts from the Journal. We also look to connect with our Knowledge Engineering work and other ongoing efforts in open science & DeSci. Please reach out if you would be interested in collaborating with us on this work in some way.
The Active Inference Ontology project reports: Over 2024 there were no major changes to the Ontology (repo). Highlights of the year included interns adding many terms and definitions, and people using the Ontology in new ways for education, programming, synthetic intelligence, linguistics, and philosophy. In 2025 we look to collaborate to make a versioned release of the ontology, with addition translations, examples, and use/application guides.
Ariel Cheng reports on a new blog post she wrote: “Things! FEP Explained, Part 1”.
ReInference
From the Symbolic Active Inference project, Jean-François Cloutier reports: I resumed software implementation after presenting current status of the project at the 2024 Active Inference Symposium. No measurements were made other than using the ability to convert design into code as a measure of the design’s cogency.
From the Active Inference Account of Belief Updating in PTSD project, Andrew Pashea, Jeremy Cooper, Haeun Sun, and Angelos Krypotos report (abbreviated here): a full initial model built, immense amount of progress on constructing and plotting hierarchical models. In any case, we have only just arrived at a working construction, so as always there is plenty of cleaning, hypothesis testing, and revising to do—fortunately we now have the needed code and logics for carrying these steps out and realizing our project!
From the An Active Inference Approach to Autonomous Navigation project, Harshil Shah, Satyaki Maitra, and Rohit Shenoy report: This project is in very early stages. So far, a project plan has been developed (and is still constantly being updated) and execution is in the early stages. This Github shows progress with code and how it evolves throughout the project. The milestone of a detailed project abstract/plan has been achieved. The team made significant progress on the first phase of implementation (a mini-grid simulation).
From the Active InferAnts project, Daniel Friedman reports: We continue to develop our open source repository, summarized in the InferAnt Stream series. We made a pull request to add Active Inference to ElizaOS. We are working to integrate Active Inference generative models with agentic mesh frameworks. We are driven this year to explore questions such as: What are the enabling conditions and requirements for symbolic coherence? What are taxonomies and application methods for spatialized and non-spatialized multi-agent models?
The RxInfer.jl learning and development project reports: Our work from 2024 was summarized at this timestep in 2024 Roundtable #4. See group document for information on weekly meetings from 2024 and through 2025. We look forward to developing more relationships with the RxInfer ecosystem, and developing the Engineering aspect of this initiative (possibly by splitting this off into another implementation-agnostic engineering project).
Support and Engage
Donate to the Institute to support our work and sustainability in the years to come (also feel free to subscribe to or share the newsletter, or contact us with other philanthropic and grant ideas).
Please get in touch if you have general or specific feedback, ideas, or questions for Institute about this year or next.
Officers