​What can natural intelligence (still) teach AI research?

​Lossfunk Research Mixers are intimate, topic-first roundtables for people in Bengaluru with a real knack for research, whether they are researchers, engineers, writers, philosophers, artists, interdisciplinary thinkers, or independent tinkerers.

​This time, we’re exploring what studying natural intelligence can contribute to AI research today and to the intelligent systems we may want to build in the future.

​Natural intelligence here is deliberately broad: humans and other animals, developing organisms, collectives, societies, and even systems such as immune networks or morphogenesis. The interesting question is not simply whether AI can imitate them, but whether they already solve important problems in ways we have not fully understood.

​How does a lifetime of experience remain useful without an ever-growing external memory store? Why do infants get to choose and shape their own training data? How can a bee do so much with so little compute? How can a swarm make decisions without a central decider? Can infrastructure adapt more like bone, or organize more like a slime mould network? What can biological policing, collective decision-making, repair, development, or cumulative culture teach us about future AI systems and the societies built around them?

​These connections may matter for AI architectures, but also for medicine, governance, law, robotics, infrastructure, collective intelligence, scientific discovery, and other domains we have not thought of yet.

​Sushrut Thorat will begin with a brief history of the interaction between natural intelligence and AI, followed by examples of where he thinks there’s still work to be done. From there, we’ll open the question to the room, drawing on the different disciplines, experiences, and perspectives people bring.

​What to expect

• ​A historical overview · 20 minutes: Sushrut’s perspective on how natural intelligence and AI have informed one another.

• ​Open questions and discussions · 15 minutes: Initial thoughts and questions from the room.

• ​Small-group discussions · 30 minutes: Groups of ~4 explore potential connections, question assumptions, and help one another refine their ideas.

• ​Sharing with the room · 30 minutes: The groups informally present the key ideas and questions that emerged.

​With participants’ permission, we’ll collect contributions into a public document. These may also form the basis of a write-up on the Lossfunk Substack.

​You should apply if

• ​You’re interested in natural intelligence, AI, or the kinds of problems intelligent systems will need to solve.

• ​You have a question or perspective from a domain you know well or are actively exploring.

• ​You enjoy testing ideas with people whose backgrounds differ from yours.

• ​You’re open-minded, willing to contribute, listen closely, and revise your thinking.

​We welcome researchers, students, engineers, founders, and people taking unconventional paths. We’ll curate a mix of backgrounds and disciplines to make the discussion fruitful. Founders are welcome to participate; this is not a pitching event.

We believe great researchers can come from anywhere. We especially encourage students, early-career professionals, motivated women, and people taking unconventional paths to apply.

​Come with a question

​As part of your application, we’ll ask you to identify something that natural intelligence seems to do remarkably well, especially under constraints that matter, and ask whether there might be a useful principle there for AI or for systems built with AI.

​It could come from neuroscience, biology, medicine, collective behaviour, law, governance, architecture, ecology, economics, or somewhere else entirely. An early thought or open question is welcome. We’d like you to think about it before coming, and we’ll help one another sharpen the good ones into concrete research questions during the mixer.

​Presented by Sushrut Thorat

​Organized by Lossfunk, a lab for time travellers.