Dr. Joel Lehman

Joel Lehman is a senior research scientist at Uber AI, where he leads the lab's AI safety research efforts. Previously, he was the first employee of Geometric Intelligence (acquired by Uber) and a tenure-track professor at the IT University of Copenhagen, where his research focused on evolutionary computation, neural networks, artificial life, and computational creativity. He was co-inventor of the popular novelty search evolutionary algorithm, and co-wrote a popular science book (featured in Harvard Business Review and 538) on what search algorithms such as novelty search imply for individual and societal accomplishment.

KEYNOTE: AI Safety for Evolutionary Computation, Evolutionary Computation for AI Safety

The talk first describes the broad aspirations of evolutionary computation (EC), and then covers two main themes at the intersection of AI safety and evolutionary computation (EC). First, what distinguishes the safety profile of evolutionary algorithms from other approaches to AI; and second, what EC techniques may uniquely offer to technical AI research.


One preliminary theme of the talk is to describe EC broadly, to clear up misconceptions about EC that might mislead those in the broader machine learning (ML) community about its risks and potential. In particular, outside perception of EC often assumes the field is the study of biologically-inspired black-box optimization. However, the most ambitious communities within EC focus not on optimization of a fixed objective, but on understanding the algorithmic nature of evolution’s divergent creativity, i.e. algorithms that are capable of continually innovating in an open-ended way.


These kinds of evolutionary algorithms (EAs) thus offer a bottom-up path towards human-level AI (HLAI), one where HLAI emerges as a byproduct of a larger open-ended creative project, as occured in biological evolution.


  • Icono social Twitter
  • LinkedIn



Jan, 2021 | Japan


© 2020 by AISafety.