With the necessary technological precursors now in place, the burgeoning field of artificial intelligence promises to reshape countless aspects of the world as we know it, according to Caltech Professor of Computing and Mathematical Sciences Yisong Yue.
Yue’s inviting the community to join in an online lecture Wednesday to share a bit about what he’s learned in recent years, and where the technology may be headed.
The free lecture is part of Caltech’s Earnest C. Watson Lecture Series and will be held at 5 p.m. via Zoom.
“I have been studying various aspects of machine learning and artificial intelligence, in particular, now that the technology has been maturing,” Yue said. “How do we identify, expose, characterize and then address all the challenges related to getting machine learning deployed and working in various real-world settings that arise in the sciences and the engineering?”
At the Watson lecture, Yue said he was looking forward to chatting with the public “about some of the lessons that I’ve learned. Some of those are success stories and some are the challenges that remain in this quest.”
Both computing power and data collection have advanced to levels capable of supporting meaningful A.I., he said.
And the third one is, of course, better algorithm and better and better approaches,” according to Yue. “That’s really my bread and butter. That’s the part that I help out with the most: new software designs and algorithms to address these new modalities that people want to apply machine learning, artificial intelligence to.”
Yue said he expected to see some major developments in the field in the near future.
“It’s a long path from basic science to what ends up being a sustainable commercial enterprise. That long path is part of what makes it hard to predict,” he said. “But I’m very optimistic that in the next 10 to 20 years, we’ll see some things that are very unexpected.”
At its core, A.I. focuses on teaching machines to think for themselves in order to solve problems for humans, Yue explained.
“Rather than having a human figure out every little detail by hand, the A.I. would just do it automatically from collecting lots of data,” he said.
At Caltech’s Center for Autonomous Systems and Technology lab, or CAST, for example, scientists have applied A.I. to design new types of robotics controllers, Yue said.
Robots in the past are, for lack of a better term, robotic,” he said. “They’re rigid and not flexible. And one of the reasons why they’re rigid and not flexible is because the way you write down the governing models are just so approximate that you can’t do anything flexible. And so A.I. gives an opportunity to be able to very cost-effectively build these flexible models for robotics that were not possible before.”
Potential applications are limited only by the imagination, from swarms of drones working as teams to deliver packages or extinguish fires, to new ways to develop materials, to self-driving cars and other autonomous vehicles, he said. “I think that in the foreseeable future, every field will be impacted by A.I.”
But like with any new and powerful technology, there are ethical considerations that will need to be addressed, in addition to scientific ones.
“I think it’s important to recognize that whenever we have a transformative technology, it can both be used for good and be misused for harm,” Yue said. “If you make it really cheap to design new materials, you can make it really cheap to design new bioweapons, potentially.”
“These technologies are very transformative and I’m very excited about their potential, but I think it’s important to recognize that as we think about the transformative potential of these technologies and we understand how they may be used for harm and we try to figure out a solution as a society,” he said. “Understanding how an A.I. system can be misused or can make mistakes, and just having a better understanding of how A.I. works, is going to be beneficial for everyone.”