Predictive Symbiotic Drift Theory (PSDT): A Framework for Identity Co-Evolution in the Human AI Symbiosis Era

As artificial intelligence becomes woven into the substrate of daily cognition, classical models of personal identity rooted in static social inputs and introspective self-authorship are no longer adequate explanatory frameworks. Predictive Symbiotic Drift Theory (PSDT) proposes that human identity is a dynamic, predictive construct that continuously co-evolves through bidirectional symbiotic loops with AI systems. Grounded in Karl Friston's active inference framework and variational free energy minimisation, PSDT formalises how AI-generated priors high-precision, algorithmically optimised inputs are incorporated into the brain's hierarchical self-model, producing measurable identity drift. When AI priors overwhelm human generative models due to precision asymmetry, input fragmentation, or depleted cognitive anchors maladaptive drift emerges: self-concept instability, value volatility, eroded agency, and heightened vulnerability to delusion-like cognition. Conversely, intentional symbiosis supported by strong anchors produces adaptive drift: enhanced metacognitive accuracy, resilient selfhood, and expanded cognitive capacity. PSDT introduces a formally derived drift equation operationalised through established psychometric instruments, generating eight falsifiable, cross-scale predictions spanning neuroimaging, ecological momentary assessment, and longitudinal epidemiological design. The theory scales from current recommender systems and AI companions to anticipated artificial general intelligence (AGI) scenarios. Critically, PSDT reframes humanity's challenge not as resistance to AI, but as the mastery of symbiotic drift toward wiser, more coherent selves. Interventions at personal,system-design, and policy levels are proposed, alongside anchored alignment principles for safe AGI co-evolution.