A New Kind of AI: Coevolutionary Hybrid Intelligence and the 4th Wave of AI Evolution

A landmark paper by HCI researchers proposes a unified framework for the next generation of AI — one that learns alongside humans, not merely from them.

Artificial intelligence has evolved in waves. The first gave us rule-based expert systems. The second brought statistical machine learning. The third — deep learning — produced the large language models and image generators that have captured the world’s attention. But a growing body of research suggests we are now on the threshold of something fundamentally different: a fourth wave of AI evolution.

A new position paper by researchers at Chulalongkorn University and CMKL University — including HCI co-founders Dr. Hossein Miri and Oleg Shovkovyy — proposes a comprehensive framework for this next stage, which they term Coevolutionary Hybrid Intelligence (CHI).

What is Coevolutionary Hybrid Intelligence?

The paper begins from a deceptively simple observation: current AI systems, however sophisticated, are fundamentally static in relation to the humans who use them. They are trained, deployed, and then used. The relationship is one-directional. Humans adapt to AI; AI does not genuinely adapt to humans.

CHI proposes something different — a class of AI systems that evolve alongside their human partners in a dynamic, mutually reinforcing relationship. Drawing on theories of extended cognition, the framework positions AI not as a tool but as an adaptive cognitive partner — embedded in human social and technical systems and co-evolving with them over time.

Why hybrid?

The ‘hybrid’ in CHI refers to the integration of two approaches that have historically been treated as competitors: symbolic AI (which represents knowledge as rules and logic, interpretable by humans) and sub-symbolic AI (which learns patterns from data, as in deep neural networks). Each has strengths the other lacks. CHI argues for their integration — systems that combine the rigour of logic with the adaptability of learning.

The ethical dimension

What distinguishes the CHI framework from previous proposals for hybrid AI is its explicit integration of ethical alignment as a core design principle — not an afterthought. The paper proposes specific architectural principles for how ethical constraints can be built into the structure of CHI systems, rather than imposed from the outside after the fact.

Why it matters

The paper’s significance lies not in any single technical contribution but in its synthetic ambition. By drawing together threads from cognitive science, AI research, ethics, and systems theory into a unified framework, it provides a conceptual map for a field that has often proceeded without one.

For the Human Continuity Institute, CHI represents both a vindication of our approach and a research agenda for the years ahead. The question is not whether AI will continue to evolve — it will. The question is whether that evolution will be guided by frameworks that keep human flourishing at the centre.

This paper is a serious attempt to build one.

Source: “Coevolutionary Hybrid Intelligence: Pioneering the 4th Wave of AI Evolution” — Paphapote, Shovkovyy, Adelifar, Chuangsuwanich, Saengtabtim & Miri. International School of Engineering, Chulalongkorn University / CMKL University, Thailand.