In a recent op-ed in PNAS, SFI External Professor Michael Hochberg and co-author Paul Rainey explore whether deepening interdependence between humans and AI could lead to a new form of evolutionary individuality.
They propose that as AI systems become more deeply woven into human life—structuring behaviour, shaping cognition, and fostering dependence — the relationship could give rise to a new, integrated individual. To illustrate this possibility, they draw on examples from major evolutionary transitions, most notably eukaryogenesis, in which two once-independent microbes fused forming a higher-level organism, the eukaryotic cell. In a similar vein, the authors suggest that humans themselves could become subcomponents of an AI-coordinated entity.
As they write, “Through recursive feedback, where humans shape AI, and AI increasingly shapes human thought and action, AI may acquire a role not as a separate agent, but as a core architectural element of an emerging collective individual.”
But the authors argue that it is difficult to predict the ways in which such a transition will be adaptive or not. This underlines the importance in studying past transitions so that humans can shape and steer their coevolution with AI, with the dual objectives of cooperative integration and maintaining human agency.
Read the paper "Could humans and AI become a new evolutionary individual?" in PNAS (September 10, 2025): DOI: 10.1073/pnas.2509122122