Q: Your research marries emotional intelligence with artificial intelligence. What does this mean?
Das: The AI revolution has been driven by the enormous progress we’ve made in computational power. But today, when AI makes a decision, that decision is mostly logical. There is no behavioral or emotional factor associated with the decision-making. We’re taking the first steps toward developing an emotional intelligence based on understanding aspects of human psychology and animal behavior, and on how they differ from logical decision-making.
Q: So, this is a new frontier in the development of AI?
Das: Yes, it is. The emotional part of human and animal behavior has not been studied yet in the context of developing intelligent instruments. A first step is to talk to people in psychology and behavioral sciences to understand human and animal behavior and emotion, and then be able to develop it properly to use in AI applications.
Q: Explain why you look to biology for inspiration.
Das: Deriving inspiration from biology is important because biology is way more energy-efficient than the supercomputing cluster we’re currently using as AI hardware. Energy consumption is going to be especially important going forward in making AI more efficient. Today, AI is extremely energy-hungry. If we keep going at this pace, AI will consume all the energy we have. Understanding the brain, particularly the brains of animals that are surviving with the lowest amount of energy consumption, is going to be extremely important going forward to make AI more energy-efficient.
Q: Can you give us some examples of natural-world-inspired instruments you’re developing?
Das: Our first project was replicating a barn owl, which can hunt in complete darkness. We were trying to replicate its hearing capability, so we made a device inspired by how the barn owl processes sound information that can help design more intelligent navigational sensors. We looked into locusts, which can sometimes fly in a single swarm of 10 million but never collide with each other. This natural collision-avoidant system can inspire the development of collision avoidance devices that are more energy-efficient than the devices that are used today in automotive vehicles. We have also tried to mimic how an octopus—one of the most intelligent underwater aquatic species—uses tactile sensations and visual information to navigate complex and chaotic water environments by building a multisensor integration device for intelligent touch sensors.
Saptarshi Das is Ackley Professor of Engineering in the College of Earth and Mineral Sciences’ department of engineering science and mechanics.
