Faculty from I2S and the Medical Center participate in panel discussion on AI

Recently, faculty from the Institute for Information Sciences (I2S) and the School of Medicine joined a panel discussion as part of the Collaborative Research Lunch series on the topic of Artificial Intelligence (AI). The panel was moderated by David Tamez, assistant research professor in the Institute for Information Sciences. Panelists included:

·      Perry Alexander, AT&T Foundation Distinguished Professor of electrical engineering & computer science and director, I2S

·      Sushant Govindan, vice chair of internal medicine, School of Medicine; and chief innovations officer, Veterans Health Administration VISN 15 Regional Office 

·      Bo Luo, H.J. and Joan O. Wertz Professor of electrical engineering & computer science

·      John Symons, professor of philosophy and director, Center for Cyber Social Dynamics 

Questions posed to the panel included topics such as what they view as bleak or not so bleak about AI in this moment, what the perils or promises they have observed in their own work, and how they are approaching AI when it comes to preparing students for the future. There was consensus that AI responses are routinely rife with factual errors or “hallucinations”. Several panelists multiple times during the discussion stressed the importance of factchecking responses.

There was some discussion about the impact AI has or will have in the future on labor markets. While there was agreement that AI will be used widely across many industries and professions, there will be a need to have a human aspect for some time to come. Alexander, whose research involves high assurance systems – or systems that if they break can be catastrophic, such as airplane control systems – said that AI technology is not even remotely close to being reliable in such systems. “We have to train ourselves to use AI technology where we are staying on top of the inaccuracies,” he said.

Symons, who grapples with AI from a philosophical perspective, talked about there being no single best general method for building machine learning systems. “Every system has to be a domain specific system,” he said. “Faculty can exercise a certain level of independence when using AI to support their work, but the challenge is that it remains to be seen if the work improves or if (AI) will degrade that work.”

As far as preparing students now for a future with AI, Govindan from the Medical Center offered his approach when working with students. “I tell them there is no point in fighting technology. You might as well learn it and figure out how you can use it productively,” he said. “The biggest thing I tell my students is that as you look at your profession, your future, what do you see that you’re going to be doing for your patients that they can’t ask an algorithm.” He also tells them that if they use AI technology, “you are still accountable for everything you sign off on. If something is wrong in your work, it is not the cause of the algorithm. You are ultimately responsible.”

The AI panel discussion took place at the Collaborative Research Lunch on Wednesday, Nov. 19, at the Kansas Memorial Union.