Is Kasparov still right?
Kasparov's conclusion: "Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process." He wrote that after two amateurs with three chess engines beat grandmasters in 2005. I kept coming back to it while watching how companies roll out AI agents today. Half of his thesis is dead. Engines stopped needing…
Garry Kasparov's stance, "Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process," is no longer valid in today's AI landscape. Chess engines no longer rely on human partners, yet the crucial aspect remains: the process determines the outcome.
A Harvard/BCG experiment involving 758 consultants demonstrated this principle, wherein the same tool and people yielded a 40% improvement within the model's range but 19 points worse just outside it. Consequently, the debate has evolved, no longer focusing on "human or no human," but rather on "who decides the appropriate level of human involvement in each use case and how can you prove its safety?"
Despite the lack of a clear answer, a framework has been developed. It consists of five autonomy levels, sixteen controls mapped to OWASP Agentic and ISO/IEC 42001, a scoring worksheet that caps the autonomy level, and promotion rules tied to evidence and revoked signals. The framework is open-source under CC BY 4.0: https://github.com/cjohannsen81/agent-autonomy-levels.
The author invites feedback from those deploying agents in production, emphasizing that the framework's thresholds require real-world testing for calibration.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
