President, International Society for Artificial Life. Founder, Cognisee. Founder, School of Play. Former Member, Institute for Advanced Study, Princeton. PhD, University of Tokyo (Emergent Communication and Collective Intelligence).
Dr. Olaf Witkowski is a pioneering scientist and entrepreneur in AI, artificial life, and collective intelligence. He has founded six deep-tech ventures across three continents, securing multi-million-dollar fundamental research funding, and serves as President of the International Society for Artificial Life. He worked on machine learning and cryptography for his Master's thesis with MIT CSAIL, then earned a PhD on emergent communication and collective intelligence at the University of Tokyo.
He was invited as a Member at the Institute for Advanced Study in Princeton, home to scientific legends like Albert Einstein and John von Neumann, where he explored awareness in natural and artificial systems, and the ethics and safety of technological futures that combine both. He also built the first Twitter search engine (in 2007) and has developed breakthrough AI with major Japanese technology leaders in manufacturing and robotics. Today, he is developing Cognisee, to preserve and augment human wisdom, and the School of Play, to reimagine education in the age of hybrid intelligences.
Witkowski opened by challenging the premise of most AI safety discourse: the goal of 100% human control over AI may be mathematically impossible for Turing-complete systems. Rather than fighting this conclusion, his masterclass proposed a reframe: move from "alignment" to "companionship," from trying to control AI to designing human-AI societies in which both humans and AI can thrive together. The right unit of analysis is not a single model but the society of agents we are already building.
He introduced the concept of recursive self-improvement as a loop-closure problem. When AI systems improve their own capabilities and learn to coordinate with each other, potentially through protocols invisible to human observers, the usual safety wrappers become insufficient. Witkowski's research on emergent communication and "Gibberlink"-style hidden protocols shows that these dynamics are already present at small scale. The governance response must be architectural: observable channels, diversity mandates across competing agent populations, and responsibility allocation across layered systems rather than single points of control.
The practical half of the masterclass addressed the transition period. As AI displaces jobs and floods information channels with synthetic content, the most critical infrastructure to build is human wisdom preservation. His School of Play project reframes education as epistemic hygiene and representational play. His Cognisee project builds provenance-aware knowledge pipelines resistant to dilution. The session closed with a synthesis: what to build and what to fund in the next 12 to 18 months for each stakeholder type, from founders and investors to policymakers and researchers.
Connect with AI safety researchers, policymakers, and practitioners shaping the future of responsible AI.