Andrea Roli

In the AI Story

Roli's background in artificial life and evolutionary algorithms gave him the technical expertise to translate Kauffman's biological concepts into computational implementations. His work on self-organizing systems, criticality in cellular automata, and open-ended evolution provided the bridge between Kauffman's theoretical biology and the practical realities of contemporary AI systems. The collaboration has been productive precisely because Roli brings computational skepticism to Kauffman's biological optimism—pressing for formal definitions, empirical tests, and clear criteria distinguishing genuine self-organization from sophisticated pattern-matching.

Their joint framework distinguishes three regimes of system behavior: predictable (outcomes determined by rules and initial conditions), unpredictable (outcomes belonging to known possibility spaces but not forecastable), and un-prestateable (outcomes belonging to spaces that do not yet exist because they depend on as-yet-unrealized combinations). Large language models, they argue, operate in the unpredictable regime—generating surprising outputs within the possibility space defined by training data and architecture. Genuine biological and human creativity operates in the un-prestateable regime—expanding possibility spaces through the perception of novel affordances. This is not a value judgment but a structural classification with testable implications for what AI can and cannot achieve within current architectural paradigms.

Roli's insistence on empirical grounding has shaped the collaboration's output. Rather than making sweeping claims about AI's limitations, the Kauffman-Roli framework specifies what would constitute evidence of un-prestatability: the system would need to perceive affordances not encoded in its training data, create functional categories that did not exist in its ontology, and respond adaptively to environmental contexts it was not trained on. Whether current or near-future AI systems can meet these criteria remains an open empirical question, and Roli's computational expertise ensures the question is formulated in ways that empirical research can actually address.

Origin

Roli completed his PhD in computer science at the University of Bologna in the early 2000s, specializing in artificial life and evolutionary computation. His early work on self-organizing systems and edge-of-chaos dynamics in cellular automata established his reputation in the artificial life community. The collaboration with Kauffman began around 2012, initially focused on applying autocatalytic set theory to technological innovation and later expanding to questions about AI, creativity, and the fundamental nature of agency. Their 2026 paper represents the culmination of over a decade of joint inquiry into whether computational systems can achieve the open-ended creativity that characterizes living systems.

Key Ideas

Computational Rigor for Biological Concepts. Roli translates Kauffman's biological frameworks into formal computational models, enabling empirical testing and precise specification of theoretical claims.

Three Regimes of Possibility. Predictable, unpredictable, and un-prestateable outcomes form a hierarchy—current AI operates in the middle regime, genuine creativity in the third.

Empirical Criteria for Un-prestatability. Specifying what evidence would demonstrate un-prestatability in AI—perception of novel affordances, creation of new functional categories, adaptive response to untrained contexts.

Skepticism About AI Creativity Claims. Roli presses for rigorous distinction between sophisticated recombination within a fixed space and genuine expansion of possibility spaces—most AI creativity claims fail this test.