Campus News

UB Council gets primer on artificial intelligence

By SUE WUETCHER

Published December 5, 2017 This content is archived.

Print
headshot of venu Govindaraju.
“AI, instead of artificial intelligence, is more like augmented intelligence, or amplification of human intelligence. ”
Venu Govindaraju, vice president for research and economic development

We’re all familiar with the sci-fi view of artificial intelligence, in which the robots try to take over Earth. But in reality, AI — “artificial intelligence for the social good” — does not replace humans, but rather augments the intelligence of human beings, Venu Govindaraju, vice president for research and economic development, told the UB Council on Monday.

“AI, instead of artificial intelligence, is more like augmented intelligence, or amplification of human intelligence,” Govindaraju said, beginning what he called a “five-minute primer” on AI. AI is the focus of a new SUNY research initiative — led by UB and Stony Brook. SUNY is asking the state to invest in the initiative.

“AI cuts through all disciplines: It’s not just about computer science,” said Govindaraju, SUNY Distinguished Professor of Computer Science who helped develop a handwriting-recognition system that has saved hundreds of millions of dollars for the U.S. Postal Service. AI touches health, transportation, education, the humanities and the social sciences, he said. “Any place where you have computing in some way, you have artificial intelligence.”

Although AI has been around for some 60 years, it’s now being called the “new electricity,” he said. “Just like electricity revolutionized several industries, AI is going to do the same thing.”

So what has changed? The abundance of data, Govindaraju said.

Data is everywhere, it’s being collected by sensors and unlike years past, when files had to be erased in order to store new data, there’s more, cheap storage space available today.

And there have been innovations in computing architecture where processors can now “crunch through all this data very fast,” he said.

Moreover, there has been the emergence of what’s called “deep learning algorithms.”

The old AI “was not scaling,” Govindaraju explained, so systems were built where computer scientists would talk to experts, pick their brains, capture all that knowledge and try to build a program that would mimic the expert. That wouldn’t scale because we can’t account for all conditions, he said.

Today, the learning is actually taking place by just giving labeled data, he said. “So a doctor does not need to teach the computer how they diagnose, as long as you give sufficient examples of good and bad scenarios. So give me the data and tell me what the doctor should have done and what he shouldn’t have done. If we have millions of such cases, the computer actually learns the mapping from the input to the output,” he said.

This technology has made AI a commodity, he noted. Whether it’s speech translation, speech recognition, object recognition, these are all available today at levels which exceed human performance.

“That’s where the excitement of where artificial intelligence is taking place,” he said.

But with this emergence of AI comes responsibility, Govindaraju said, citing civil liberties, privacy, cybersecurity and mass unemployment among areas of concern.

“These are exciting times and we need to address these issues,” he said.

Govindaraju pointed out that UB has a “rich history” of work in artificial intelligence, noting UB’s handwriting-recognition system has been called the “first major success story” in the field.

The university is continuing that important work, with scientists conducting research on a variety of AI applications, among them self-driving cars, advanced manufacturing, precision medicine and digitization of legacy documents, he said.

He noted that when he called a meeting in August to begin preparing for the SUNY AI initiative, 160 faculty members expressed interest. For that initiative, he said, UB is working on large-scale proposals in the areas of health informatics and autonomous systems.

In other business at Monday’s meeting, President Satish K. Tripathi told council members that UB continues to work with the Western New York congressional delegation, the Association of American Universities and the Association of Public and Land-grant Universities to ensure that provisions that would negatively impact UB students are not included in the final federal tax reform bill that lands on President Trump’s desk.

The bill passed by the House Nov. 16 would eliminate the current federal tax waiver on college and university tuition waivers provided to students who serve as teaching or research assistants while pursuing advanced degrees.

Graduate students at UB and across the country say including tuition waivers — UB uses the term “scholarship” instead of “waiver” because UB still must fund the tuition to SUNY —as taxable income would dramatically increase their tax burdens and could affect their ability to stay in school.

The House bill also eliminates the tax deduction for interest paid on federal student loans. Under current law, borrowers can deduct up to $2,500.

While the tax bill passed by the Senate on Dec. 2 does not include either of these provisions, Tripathi says UB will continue to work make sure they do not appear in the final version of the bill that will be worked out in conference committee.