Amnon Shashua
While his peers pursue the creation of generalized intelligence, Shashua’s betting that by narrowly focusing on superintelligence in STEM, his lab may be able to produce “great scientists at scale” within the next few years. Though current models fail to engage in deep reasoning, he says, AAI has proven a “new training methodology” that will one day allow it. Will this superintelligent AI be aligned with our values and goals? Shashua believes, for mathematical reasons, that is impossible. “I don’t think this is a problem that will be solved,” he says. “It’s a problem that will be managed.” Still, he believes it’s worth pursuing.
When it comes to autonomous driving, Shashua says the way forward is clearer—there are challenges, but no uncertainties. Mobileye has signed deals with Uber and Lyft to roll out autonomous robotaxis starting mid-2026. On the robotics front, though, uncertainties remain. Shashua expects that humanoid robotics will be able to reliably perform “pick, drop, and move” tasks by the end of the decade.
Tharin Pillay