Shaping the Future: Aerospace Enters Breakthrough Artificial Intelligence Competition
Aerospace’s Dr. Terence Yeoh and Dr. Nehal Desai are competing in a three-year cognitive computing challenge that promises to accelerate the understanding and adoption of powerful artificial intelligence (AI) technologies.
The IBM Watson AI XPRIZE competition aspires to show how AI and humans can collaborate to create advanced applications to benefit humanity. In fact, entrants are encouraged to push the boundaries of human potential by focusing on problems that are believed to be unsolvable.
The creators of the competition believe that the field of artificial intelligence will likely be this generation’s greatest achievement, similar to the Internet and personal computing.
Yeoh and Desai are taking on the grand challenge of designing an AI that can dream.
“At Aerospace, we have always applied innovation for the good of the nation, and we thought this was a wonderful opportunity to apply innovation – the DNA of our company – to something that would benefit humanity,” Yeoh said.
The teammates are leading this effort through SeedTECH, an after-hours community of interest group that brings volunteer Aerospace scientists and engineers together to develop technology that is aligned with the corporation’s interests.
How would their entry push the boundaries of human potential?
The team claims that AI to date is used for assistance, and they are looking to create AI that is more of a colleague to bounce ideas off of or to come up with new ideas.
“The key thing here is to open up the aperture and to be able to look for solutions that are beneficial for everyone,” Yeoh said. “Our challenge is to derive an algorithm that is creative, rather than reductionist, by showing increasingly new associations that are resilient to overtraining.”
“Today’s AI systems are being applied for the purposes of efficiency, but we believe the ‘why of AI’ should be altruism – to design systems that benefit the greater good of all,” Desai said.
He explained that dreaming plays an important role in memory consolidation within the human brain. The ability to create, imagine, and daydream, he said, is linked in some way, for example, to the ability of a child to look at only one or two images of a cat, and through this means, instantly recognize all cats.
“In contrast, artificial neural network-based systems require tens of thousands of images to recognize a cat with high certainty, but can be vulnerable to overtraining and misclassifications,” Desai said. “By deriving a new artificial intelligence system from the ground up that is creative in the learning process rather than reductionist, this system provides solutions that will enable higher-order operations that are not achievable with the current reductionist approaches.”
The team embodies the “fail successful” philosophy of the SeedTECH COI they lead, as they pave the way for future teams to compete in similar competitions.
“Aerospace has such a supportive environment – we feel like we are in a startup incubator,” Yeoh said. “Where else would you have legal, insurance, and media divisions to help with running the team, enabling us to just focus on the technology development?” he said.
The team is preparing to submit round one results on Sept.1. There are two more rounds of competition after the first round this year and the grand prize international competition will take place at the 2020 TED Talks with a prize of $3 million.