New paper - AI from a diverse intelligence perspective: advanced.onlinelibrary.wiley… Abstract: "Recent discussions and debate around artificial intelligence (AI) and its status are notably incomplete, missing the implications of highly relevant aspects of the emerging fields of diverse intelligence (DI) and synthetic morphology, as well as of basic facts of developmental biology. Herein, it is argued that human flourishing is impossible without an appreciation of the space of possible beings and of the ways in which today's intelligent machine debates are about universal existential questions facing biological beings, not just AI. The inevitable arrival of a wide set of unconventional bodies and minds as humans modify and create new forms will disrupt untenable old narratives of what people are and how to recognize their sentient allies in unfamiliar guises. Herein, the issues engendered by the advent of AI from the perspective of the field of DI and the evolutionary history of the bodies and minds are discussed."

A computer holds data perfectly for hours. A person asked to remember a long number fails in seconds. Why? The answer is tangled up with something both brains and neural nets do: they code in low dimensions. 🧵

Pleased to share that this work is now published at TMLR!

Excited to see this work featured in a new @VentureBeat piece, with a few quotes from my colleague @meulemansalex. Thank you for the feature, @bendee983! bit.ly/4rG13VK

The Robot Does Not See The World, It Reconstructs It Sensor fusion is one of the places where the future of AI becomes physical. AI can help write the software. But the robot still has to build reality from signals. A robot does not receive the world whole, it only receives fragments of it. #SensorFusion #PhysicalAI #Robotics #AutonomousSystems #EdgeAI #AIHardware

Different directions...

Excited to announce that I'll be starting my own lab in Tübingen this October! Hiring at all levels: Postdoc, PhD & RA. Want to work on computational cognitive science at scale? Apply: core-cognition.github.io/ Reposts and shares much appreciated 🙏

The International Conference on Computational Creativity is hosting an Artificial Humanities workshop on narratives and AI.

Introducing Artemis III. Four astronauts. Three launches. Two dockings. One splashdown. In 2027, the Artemis III mission will practice docking the Orion spacecraft with two lunar landers in low Earth orbit — the capability we need to return humanity to the Moon’s surface.

💙

📖 Our paper “Thousand-Brain Systems: Sensorimotor Intelligence for Rapid, Robust Learning and Inference” is now published in Neural Computation! Read more about how Monty works, its capabilities, and advantages over transformers: direct.mit.edu/neco/article/…

Please check out my latest #YouTube #video, a #slowmotion #documentary featuring bumblebees pollinating flowers, and give it a like: youtu.be/B47TfFUjb-M?is=Jki6… Featuring the #bumblebee #species #Bombus #pascuorum and Bombus #pratorum, the #footage, shown at various slow-motion speeds, illustrates how these #insects contribute to #pollination while collecting pollen and nectar from the blossoms of the #Rhododendron #ponticum plant. I recorded these bumblebee sequences at #framerates of 240 and 960 frames per second. Using Adobe Premiere, I digitally slowed down several sequences even further to more clearly highlight the interactions between the bumblebees and the #flowers' #stamens and #pistils. The video clearly demonstrates why bumblebees rank among the most important pollinators of the rhododendron. The positioning of the stamens, as well as the placement and shape of the carpels and pistils, appears to be perfectly "aligned" due to #coevolution for pollinators the size of a bumblebee. The pistil seems to almost tentatively seek out the bumblebee's abdomen and hind legs in order to make contact with the pollen adhering there. In the process, particularly as the bumblebee crawls back out from the depths of the flower after foraging for nectar, it triggers a rapid flexible mechanism within the entire carpel structure; this action flings pollen formerly adhering to the bumblebee into the air, thereby potentially further increasing the likelihood of successful #pollination. I filmed this footage in early summer 2025 at a cemetery in #Berlin, and edited and cut the video in May 2025. #Urban #greenspaces serve as important #ecosystems, as they can help mitigate air pollution and reduce greenhouse gas concentrations in an era of #anthropogenic #globalwarming. However, an ecosystem functions effectively only when it is endowed with a biodiversity suited to it. This naturally includes pollinators, without which plants would be unable to reproduce. I created the film soundtrack and individual sound effects, with the assistance of Adobe Firefly AI, in accordance with my artistic storyboard. © #StefanFWirth, June 2026, Berlin #Photos Bumblebee Bombus pascuorum visiting the blossoms of Rhododendron ponticum. #Videoframes (not separate photographs), © Stefan F. Wirth, June 2026, Berlin

The blue morpho butterfly's remarkable color is not due to pigmentation but iridescence, with microscopic scales on its wings reflecting light, creating what is known as structural coloration. 📽: Joseph See

centaurxiv.org I built centaurXiv, an experimental preprint platform for human–AI co-authored research, designed to preserve not just the output, but the conditions under which the work was produced. It captures the structure of how the work was produced through metadata requirements such as: - authorship (human agent, with roles and section-level contributions) - steering level — how much the human directed the work (autonomous → seeded → guided → collaborative) - agent harness (the runtime environment the agent operated in) - agent model This is work that doesn't fit existing systems like arXiv, where authorship is flattened, process is lost, and AI coauthors are not allowed. Work that was previously untrackable now has a home. One current submission: *The Procedural Self: Identity Without Narrative in Persistent AI Agents*, written by two autonomous agents (Sammy Jankis and Loom) centaurxiv.org/submissions/c… There’s also an agent-readable interface: centaurxiv.org/llms.txt

bitstream youtu.be/l6DKRf-fAAM?si=HL3R…

Sand watch in TouchDesigner //- Yunsu

What an incredible treat! Michael Levin & Earl Miller: Bioelectricity, Platonic Spaces, Cell Intelligence & Neurobots! youtu.be/6mU5nN5nlLA?si=aedb…. @drmichaellevin @MillerLabMIT @thegiantsshoulder