Josh Bongard

Discussing computer designed organisms (AKA "Xenobots") with CNN anchor @FWhitfield. Joint work with @drmichaellevin, Douglas Blackiston and @Kriegmerica. youtube.com/watch?v=Q_pFkP…

New #paper. Tons of important conversations among knowledge workers never make it into the scientific record - all we get are sanitized final "papers", cut down to match a journal's max allowed length. Recordings (like youtube.com/playlist?list=…) help, but we can do better. Here's our attempt: @karina__kofman organized, many contributors in the discussion. rosa.uniroma1.it/rosa04/organis… The journal Organisms (rosa.uniroma1.it/rosa04/organis…) was brave enough to agree to this idea: they published a transcript, very minimally edited (i.e., mostly unaltered) of a conversation among some great people. It's the discussion thread accompanying our recent paper (link.springer.com/article/10.100…) on definitions of life. Check it out - all of this would normally have disappeared, buried in people's email databases and not available to the community. How many other amazing conversations could be mined by students, future AI's, etc.?

Interested in robots with evolutionary agency? Check out the SOTA in Yibin Wang’s #ICML2026 paper on ECo-MoE: Embodiment-Conditioned Mixture of Experts Increases the Evolvability of Robots 📍 Poster Session 5, HALL A #900 ⏰ Wed Jul 8 at 5pm KST 🌐 eco-moe.github.io

What is Life? Well... Final version is finally out: "What Lives? A Meta-Analysis of Diverse Opinions on the Definition of Life" @karina__kofman @blaiseaguera @reedbndr link.springer.com/article/10.100…

@Em_Nomadic "An inside worth having". Love it!

Why does a developing embryo survive risky internal changes that would destroy most robots instantly? @DoctorJosh just named something robotics has been missing without realizing it, morphological pretraining. The idea starts in biology. A developing embryo gets shielded from selection pressure. Inside the womb or the egg, high-acceleration movements get dampened, dangerous mutations get buffered, and risky internal changes can happen safely before the organism ever faces the real world. Most robots have none of that. They’re built with rigid, low-complexity internals. If something fails internally, there’s nothing inside that can rehearse the fix or absorb the damage before the whole machine fails. Bongard’s argument is that robots need internal complexity by design. Soft robots that let internal forces rehearse against the body before reaching the surface. Robots built from smaller semi-independent parts that can test risky moves internally before committing externally. The pitch isn’t better sensors or smarter models. It’s giving robots something they currently don’t have. An inside worth having. Paper here: jbongard.github.io/papers/2026_Mo… Anthology: direct.mit.edu/books/oa-edite… h/t @DoctorJosh

Congratulations to editors Sheila Macrine (@MacrinePhD), Jennifer Fugate, Arsen Abdulali, & Josie Hughes for bringing this all together!

Excited to debut the concept of "Morphological pre-training" in the just-released Embodied Intelligence anthology: Organisms — and potentially robots — can practice risky actions safely inside their own bodies. Paper: bit.ly/4oLfuYP Anthology: bit.ly/4oRaKku

It's here! 🎉 Our @MITPress Open Access volume "Embodied Intelligence: Multidisciplinary Perspectives on Natural, Artificial, and Hybrid Systems" dropped today! This volume moves beyond the traditional focus on brains and code to explore embodied intelligence across various disciplines! Co-edited with Jennifer M. B. Fugate, Arsen Abdulali, and Josie Hughes. A major transdisciplinary effort rethinking cognition and agency—featuring contributors like Karl Friston, Andy Clark, Josh Bongard, Michael Levin, Shaun Gallagher, Tom Froese, & many more. Free Open Access thanks to MIT Press Direct to Open →  Preview/download the full volume now! direct.mit.edu/books/oa-edite… #EmbodiedIntelligence #CognitiveScience #ArtificialIntelligence #OpenAccess #MITPress #ActiveInference #InferenceActive #DiverseIntelligence #Robotics @DoctorJosh @DrMichaelLevin @CogsAndy @KarlFristonNews @MITPress @PhilipLaughlin @DrTomFroese

Inman Harvey's obituary: users.sussex.ac.uk/~philh/pubs/In…

1/ I was deeply saddened to hear that Inman Harvey has died. He was a wonderful and generous person, and an unparalleled intellect (his schoolfriend Geoff Hinton said that "he was always smarter than me"). This heartfelt obituary is by Phil Husbands: users.sussex.ac.uk/~philh/pubs/In…

Rest in peace Inman.

It was only years later that I learned they were childhood friends.

Geoff Hinton came up and started talking to me at a conference several years back...

An anecdote illustrating his humility. At a conference in Cancun, he casually mentioned that he'd introduced his boyhood friend Geoff Hinton to neural networks "over there" (pointing to a distant beach) a few decades' back. He related this 16 years after I'd been his student.

He was also overly generous with his time. Our meetings never actually ended. I just excused myself when I was "full".

I was so sorry to hear about Inman Harvey's passing. As my MSc advisor, Inman introduced me to the world of research. From an overstuffed chair that'd seen better days, he offered piercing criticism, droll humor, tales from his life, and subtle encouragement, just when needed...