Transcending natural evolution in the age of artificial intelligence

Artificial intelligence (AI) is undergoing a fundamental transition from digital representational models to embodied, physically situated systems. This position paper characterises this shift as the crossing of an embodiment threshold, where Al’s operational capacity becomes integrated with physical environments and neural substrates. We argue that this transition is defined by a profound evolutionary tempo mismatch between rapid algorithmic iteration and the millennial-scale inertia of biological evolution. By synthesising advances in flexible bioelectronics and high-bandwidth neural decoding, we examine the biophysical mechanisms that could enable human-machine symbiosis while exposing the thermodynamic and material constraints that accompany this trajectory. In particular, we contrast the brain’s highly efficient metabolic architecture with the resource-intensive scaling of silicon-based computation, highlighting the risk that the infrastructure required for large-scale augmentation may produce durable socio-technical stratification. We therefore contend that the central challenge of this era lies in governing neuro-computational co-evolution rather than merely regulating standalone AI systems. In this context, governance is not merely a regulatory constraint, but a deliberative mechanism for shaping whether, how, and under what conditions deeper human-machine integration should proceed. To address this shift, we propose an adaptive governance framework centred on infrastructural legibility, operational safety, and neural privacy. This roadmap outlines a strategic architecture for navigating the transition towards secure and equitable forms of human-machine symbiosis.