OB Research Seminar - Maria Pykälä, HEC Lausanne

A balance of exploration and exploitation, or individual and social learning, shapes human cumulative cultural evolution and innovation. Today, algorithms can increasingly explore and combine information from vast problem- and cultural landscapes in a manner that far exceeds human capacities. The impact of such exploration on human innovation and cultural evolution is unknown, and should be influenced by how humans learn from algorithms. This project investigates the impact of algorithm search on human cumulative cultural evolution using experiments and a simulation. Simulation results show that a balance of human exploration and learning from powerful algorithms can augment innovation. On the other hand, extensively searching, opaque algorithms can exceed the learning capacities of humans and limit innovation. In the experiment participants were in human-only or hybrid human-algorithm teams to solve an innovation task. Results suggest hybrid teams perform equally well to human-only teams, and that the presence of an algorithm and information about its behavioural rules can influence human learning and exploration behaviour. The scale of algorithm search relative to humans, learning biases and a balance of human exploration and machine exploration should influence how algorithms shape cumulative cultural evolution.