IBM director of research explains how AI can help companies leverage data

Dario Gil, IBM senior vice president and director of research, wants you to think about artificial intelligence differently.

While large language models like ChatGPT and their ability to converse may get much of the attention, Gil says the real value of AI is in the unprecedented amount of data that it can analyze and what we as humans then do with that analysis. 

“Don’t be an AI user, be an AI value creator,” Gil told an audience of UB faculty, students and stakeholders Friday on the Downtown Campus. “Use the technology but think more structurally about how to create value in whatever domain you are in, whether you’re a professor or student or an institution. A lot of this is going to be taking the data in your fields and going on this journey of unlocking its true value by embracing the power of this new representation and fully exploiting what it gives us.”

Gil gave the keynote address of the sixth edition of the UB | AI Chat Series, titled “Exploring the Future of AI for Maximum Industry Impact.”  

The event included a panel featuring Gil and other officials from some of the nation’s leading businesses. Moderated by Kemper Lewis, dean of the School of Engineering and Applied Sciences, the panelists discussed the ways their companies are currently using AI and how they’ll use it in the future.

A clear consensus emerged: AI will help companies take full advantage of their data. 

Gil, who directs IBM’s strategies in AI, semiconductors and quantum computing, explained how AI, neural networks in particular, transform our information — documents, images and even speech — and transform them into what’s known in the AI field as tokens. Foundational models can intake tens of trillions of these tokens, worth many terabytes of data.

“When you actually represent information in that fashion, you can establish semantic connections [much easier]. In this high-dimensional space, the distance between those representations tells you something about how connected those pieces of information are,” Gil said. “While we can only think in three dimensions, the machines don’t care how many dimensions there are. So it turns out to be a very powerful way to establish connections.”

Another panelist, Chris Tolomeo, senior vice president and head of banking services for M&T Bank, says the company is already using AI to confirm new customers’ identities and catch check fraud.

“I think the promise of AI will simplify things to make it much easier for both our customers and our employees,” he said.

However, he also offered caution.

“There’s high expectations that things will happen very quickly, and history tells us technology tends to be very difficult and expensive,” he said. “Not to downplay the potential, but making sure we’re doing things with the right expectations.”

AI will also transform health care and research. Allison Brashear, vice president for health sciences and dean of the Jacobs School of Medicine and Biomedical Sciences, envisions AI conducting predictive modeling for who is at risk for Alzheimer’s disease or a heart attack.

“That’s the aspiration here: Go from fixing people after they are sick to fixing people before they are sick,” Brashear said.

With UB recently being named the home of Empire AI and its supercomputing center, Gil also offered some insight on what the university can expect. IBM just recently built another supercomputing center for AI in Texas. 

“You’re going to go on this journey with Empire AI of creating foundational infrastructure that will allow people in their own fields to go and create their own foundation models or build from ones in the community and expand it,” he said. “This is going to be a collaborative and community endeavor, and of course to do this, not only do you need to have data and expertise, you need a lot of computing power.”

Last month, IBM open-sourced its series of AI foundation models, Granite. 

“The future of AI is to be open. This is an indispensable element to make it safe, innovative and distribute economic benefits to the world,” Gil said. “It is indispensable for the future of universities. If we go down a path where AI becomes closed … it’s going to be a real catastrophe for universities.”