Neil Jacobstein 2019 AI Update #suglobalsummit

Neil Jacobstein just gave a talk on Artificial Intelligence at the 2019 Singularity Summit.

He described the advance in robotics and artificial intelligence.

He reviewed the number of training steps needed for deep learning systems like Alpha Go Zero and Alpha Star to surpass human champions or the best prior software.

The number of training steps were in the 100,000 to 400,000 range.

Neil has rules that companies should follow to select good AI projects and advises using the online courses to train staff.

Neil wrote a paper in Science. Autonomous vehicles: An imperfect path to saving millions of lives. He indicates that although self-driving cars are a long way from being perfectly safe they will steadily get safer than human drivers.


Neil discussed how the Deepmind AI software, Alphastar, beat human players in the strategic computer game Starcraft.

Exploration-Exploitation Balance: In StarCraft II there is no single winning strategy. At any given time, the AI agent needs to balance the need of exploring the environment in order to expand its strategic knowledge instead of taking actions that can yield immediate benefits.

Here is a prior talk by Neil Jacobstein.

1 thought on “Neil Jacobstein 2019 AI Update #suglobalsummit”

Comments are closed.