For years, the collective imagination about artificial general intelligence (AGI) has been dominated by the figure of the “genius model”: a monolithic, superintelligent and self-aware system, capable of solving any problem from a position of cognitive superiority. This vision, inherited from a technological tradition centered on the accumulation of capabilities at a single point, is beginning to be questioned by an emerging line of researchers who propose a different paradigm: that of AGI as a distributed economy of agents. A recent article by Sébastien Krier and Benjamin Irving, “Distributional AGI Safety”, published under the aegis of Google DeepMind, which I discovered thanks to Professor Tyler Cown, suggests that general intelligence cannot come from a single… See more
About The Author
