Position: Human Evolution Should be Treated as a Goal in Human-AI Co-Evolution

As AI evolves from a chatbot that answers questions into a fully autonomous agent that can run entire workflows, humans must also evolve to adapt to this changing landscape: from judging how to use its answers to overseeing the systems that AI operates. Yet existing research on human-AI co-evolution remains AI-centric, focusing either on AI self-evolution or on involving only AI researchers, largely overlooking the evolution on the broader human side. In this position paper, we argue that human evolution should be treated as an explicit goal of human-AI co-evolution. If humans fail to co-evolve, we suffer from uncaught AI errors, degraded human feedback compounds those errors in future models, and institutional accountability erodes. To make this position actionable, we propose a four-phase framework grounded in prior work on automation, spanning humans' use of AI as a Tool, Assistant, Executor, and Orchestrator. For each phase, we analyze why human evolution matters and offer recommendations for both sides: what humans should do to evolve, and how AI design can support this process.