a scholastically deep talk at @iclr_conf workshop, by @Yoshua_Bengio on designing Scientist AI (honest, non-agentic AI) as a safe building block and guardrail for agentic AI (a talk with similar contents at @SimonsInstitute : youtube.com/watch?v=hybMno7h…)

The main idea: use a consistency loss! (3/3) joint work with Zhong Yi Wan (Google), Leonardo Zepeda-núñez (math.wisc.edu/staff/zepeda-n… ) and Anudhyan Boral( @bloopsie )

How do we parameterize the vector fields with a smoothly varying representation? (right). (2/3)

Simulating wildfires before they happen youtu.be/XEbbJHFwkUo - work from our teams at Google research (@GoogleAI ) on applying ML technology to solve computational problems and to deliver meaningful impacts.

By Claude Shannon in 1956

and here is mine - I think of renaming him sonata or atonal !

Yoshua Bengio and Daniel Kahneman discussing causation, system 1 and system 2 #aaai20 @RealAAAI (@ArielConn )

with @Mark_Mitton @frossi_t and @conitzer we had put out a fantastic show of magic demonstrating the role of attention in human cognition #aaai20 @RealAAAI - hats off to the master !

Another great moment #aaai20 @RealAAAI with Grand Master @Kasparov63

fireside chat with Danny Kaheman in system 1 and system 2 in psychology and AI - an interesting comment about Transformer too..#AAAI20