UMBC is exploring the integration of computing across disciplines, and research, such as exploring how artificial intelligence could make sensors more useful.
While the first wave of AI was about what the systems could do, the next will be about how, where, and who they do the work for. That shift points toward three big goals: making AI more efficient, more physical, and more human.
At UMBC, those goals are shaping research and teaching that bring computing together with other disciplines, including work on how AI can make the world’s sensors far more useful.
Creating efficient systems
As AI systems become more widespread, so does their need for energy and computing power. Researchers are now working to make them less energy-intensive and more efficient so that AI can run on more devices, cost less to operate, and grow without straining resources.
This work draws heavily on foundational ideas in mathematics, algorithms, and computational efficiency, and the kind of thinking that asks not only whether a system works, but how well.
“Computing is a tool of inquiry across disciplines,” says Anupam Joshi, UMBC’s vice provost and chief AI officer. He emphasizes that AI skills can prepare future workers for careers extending well beyond the technology industry, in fields such as medicine, energy, and manufacturing, because “computers are used everywhere.”
AI in the physical world
Much of today’s AI lives in software. But, a growing area of research is about embedding it in physical systems, such as robots and autonomous vehicles, where it has to sense, decide, and act in real time. The stakes are quite different when AI moves through the physical world; systems must cope with unpredictable environments, respond quickly, and work reliably alongside people.
Roberto Yus, an assistant professor for UMBC’s new Master of Science in Artificial Intelligence program, is using AI to improve the usefulness of the billions of internet-connected devices and sensors—such as video doorbells, automated thermostats and motion sensors—that are deployed around the world and are known as the Internet of Things.
A central challenge in the Internet of Things is the gap between the raw data collected by sensors and the higher-level questions people want answered, Yus says. A school principal, for example, might want to know whether classrooms have adequate ventilation while students are present. A hospital nurse might need to locate available wheelchairs. A homeowner might want to understand whether a house is energy-efficient and where the energy is being wasted.
Rather than requiring people to identify specific sensors or understand how data is collected, Yus envisions a system that automatically determines which sensors to consult, which AI models to apply, and how to combine the results to answer a question.
AI that works for people
Even the most capable AI is only useful if people can work with it. That’s why researchers are also studying how people interact with technology: how we ask it questions, how much we trust it, and where it fits into daily life and work.
Education can start with those everyday questions. At UMBC, Yus challenges students to identify problems on campus that technology could help solve. For example, in one class, they are thinking about how to check the line at a popular dining location before walking over. Handson activities Yus oversees at two Maryland high schools and a summer camp for Baltimore-area students will also give young people opportunities to build systems that address challenges in their schools and communities, helping them envision themselves as future scientists and engineers.
Preparing for an AI-driven workforce
What do these developments mean for people hoping to establish or expand their careers while keeping up with an AI-driven workplace?
Remember the fundamentals; understanding math, science, and computing are crucial components of what lies ahead. But, so is understanding people. An appreciation for where science intersects with psychology, design, and ethics will be just as important for the coming chapters of AI.
To help, UMBC has created an M.S. in Artificial Intelligence that goes beyond the fundamentals or what you could expect from a general computing degree. It is a focused, rigorous pathway into AI research, development, and deployment for students with a computing background. The program responds to demand from labs and industries in Maryland and across the region that are focused on AI.
The next chapter of AI will be written by many people, and shaped most by those who can think futuristically across what will make AI most useful: systems that are efficient, capable of working in the physical world, and simple to understand and trust.