About

Communicative Machines (CMLabs) was started in 2006 by a group of artificial intelligence and software experts from M.I.T., Edinburgh University and elsewhere, bringing together seasoned professionals across AI, robotics and computer graphics.

For the longest time futurists assured us that intelligent machines would be commonplace by now. A vast number of research projects in A.I. are started and finished each year, many of them looking for the golden screw — or algorithm — that makes the dream come true. However, research in psychology and brain science indicates that the key to intelligence is unlikely to hinge on a single principle or two. To be sure, work in cognitive science and A.I. to date has certainly increased our understanding of various aspects of mind. But the results are for the most part an unrelated collection of tidbits — not the full, holistic picture that would explain the workings of a normal brain, animal or human. Researchers continue to divide-and-conquer, instead of integrating, and as a result the work has benefited far fewer real-world applications than we would have wanted or expected, and failed to explain to a satisfactory level how a mind emerges from the workings of a brain.

The future of A.I. lies not in a golden algorithm, or two, but in various golden combinations of highly diverse techniques and technologies.

That thesis is exactly what Psyclone was built to prove. Rather than waiting for one model to do everything, Psyclone combines fast, deterministic control with slow, powerful reasoning inside a single real-time runtime — the architecture the industry now calls the “System 1 / System 2” or fast/slow bridge. Today, the golden combination we set out to build is what we now call Sub-Millisecond AI.

Original Co-founders

CMLabs grew out of the work of two researchers who had spent their careers building integrated, real-time AI systems rather than isolated algorithms.

Kristinn R. Thórisson, co-founder of Communicative Machines

Dr. Kristinn R. Thórisson

Dr. Kristinn R. Thórisson is an artificial intelligence researcher and Professor of Computer Science at Reykjavik University, where he co-founded the Center for Analysis and Design of Intelligent Agents (CADIA). He earned his PhD from the MIT Media Lab in 1996 with the thesis Communicative Humanoids: A Computational Model of Psychosocial Dialogue Skills, and is founder of the Icelandic Institute for Intelligent Machines (IIIM). A leading proponent of constructivist AI methodology and unified theories of cognition, he leads work on AERA (Autocatalytic Endogenous Reflective Architecture) — a self-programming system that learned multimodal spoken interaction purely by observation in the EU-funded HUMANOBS project. His architectural work has contributed to interactive humanoid systems, including dialogue technology for Honda’s ASIMO robot, and his research on artificial general intelligence has earned multiple awards at international AGI conferences.

Thor List, co-founder of Communicative Machines

Dr. Thor List

Dr. Thor List is co-founder of Communicative Machines and the creator and architect of Psyclone AIOS, a real-time distributed middleware for building large integrated AI systems. Holding degrees in physics and computer science from the University of Copenhagen, a master’s in astrophysics, and a PhD in artificial intelligence from the University of Edinburgh, he was a research fellow at Edinburgh’s Institute for Perception, Action and Behaviour, where he worked on the EC-funded CAVIAR computer vision project and introduced the Computer Vision Markup Language (CVML) and plug-and-play vision architectures. He led the mBrane real-time distributed runtime in the EU FP7 HUMANOBS project, and his engineering work spans humanoid robotics — including the Cognitive Map architecture for Honda’s ASIMO — computer-vision platforms, and real-time voice AI.

See where it’s headed

The thesis we started with ships today as Sub-Millisecond AI. See where it’s headed →