Stony Brook University experts explored the groundbreaking technologies reshaping our interactions, and delved into the dynamic intersection of innovation, creativity, ethics and artificial intelligence in a webinar titled “AI Horizons – What Does Innovation, Creativity, and Ethics Mean in the Generative AI Era?”

The event was a collaborative presentation between the Center of Excellence in Wireless and Information Technology (CEWIT) at Stony Brook University and the Institute of Electrical and Electronics Engineers (IEEE-USA).

Panelists included Steven Skiena, a distinguished teaching professor in the Department of Computer Science and an expert in algorithms, data science and natural language processing; Margaret Schedel, a professor in the Department of Music and a leading expert in electronic music, sound design and human computer interaction; and Vivian Zhang, chief technology officer at SupStat, a statistical consulting company. The discussion was moderated by Laura Lindenfeld, executive director of the Alan Alda Center for Communicating Science, and dean of the School of Communication and Journalism.

The panelists, representing three different disciplines, described how AI is affecting the innovation and creativity related to their respective fields.

Steve Skiena

“I’m in the creative arts, so creativity is built into what I do,” said Schedel. “I have spent decades exploring the intersection of human creativity and what was then called ‘machine learning.’ Now the umbrella term is ‘artificial intelligence.’ My work focuses on creating systems that enhance and interpret human creative expression, rather than replacing it.”

“When I thought of innovation and creativity, that usually meant developing new technologies, new algorithms, and new systems,” said Skiena. “Today what is considered research and what is considered creativity is changing. A lot of what my students are doing these days involve using models rather than building everything from scratch.”

“I’ve been at the frontier of building algorithms and startups,” said Zhang. “One of our startups helps dentists submit their claims to avoid rejection from the insurance company. In fact, dental offices spend about 40 percent of their operation costs on just getting the insurance approved. Creativity to me is navigating that.”

Skiena said that the current large language models (LLM) enable all kinds of creative endeavors that would otherwise not be possible. To make the point, he described a project in which his class rewrote The Great Gatsby without using the letter “E” anywhere in it.

“It’s perfectly readable and tells the same story,” he said. “My students and I programmed the LLM in the right way, and it’s phenomenal. That idea is simple, but it would have been impossible to do without these models. It’s exciting that these strange ‘dream ideas’ can now be done relatively easily.”

Schedel described three paradigms: AI working alone, humans working alone, and the integration of humans and AI.

“AI really excels at these combinatorial tasks and pattern recognition, but it lacks deliberative, goal-oriented thinking,” she said. “Humans bring this intentionality, this emotional understanding, this embodiment, and the ability to set these meaningful, creative goals. That synthesis creates exciting possibilities for me.”

Margaret Schedel

“I see new applications happening every day,” said Zhang. “My challenge is connecting the dots to get problems solved.”

Schedel described the impact AI has had on photography.

“AI photography is entering competitions. Is that not okay?” Schedel asked. “The people inside are saying, ‘Maybe we should have separate competitions for these AI-generated things.’ And maybe we can call them ‘promptography,’ because you put the prompt in and it isn’t as simple as just saying ‘make a cool image.’ The more detailed your prompt is, the more creative the output becomes.”

“In some sense, as these tools become easier to use, that level of creative magic presumably becomes less important,” said Skiena. “But it’s still a question of what image do you want to see? That feels like that’s got to be creative on some level to me.”

The webinar concluded with a question from the audience: “How will artificial intelligence change education?”

“Does the change level off to the point where one can think intelligently about how education should change?” Skiena asked. “It’s hard to think about that while everything is changing. In my heart of hearts, I believe that knowing how to write and think is important. I believe that understanding math and how to reason is important. I believe that having the knowledge to compare things and assess sources and have some background for reasoning is important. And I’d like to think this will still be true even as machines become smarter.”

“From what I have seen, homework and exams need to be completely redone,” said Zhang, who taught at Columbia University for 10 years. “Students use AI to come up with some initial question and work upon that. So the current homework and exam methods won’t work anymore. A new way of challenging their thinking process and the learning outcome needs to be established.”

Vivian Zhang

“I believe they’re coming up with solutions faster, but the question of whether they’re learning faster is less obvious to me,” said Skiena. “There’s something called Metcalfe’s Law, which says the power of a network grows quadratically with the number of nodes in it, because you have more things, the more you can react to, and the more valuable it is. But it’s important to me that people still be able to retain things and understand outcomes from these models. That needs to be done from a background of skills and a background of knowledge.”

“I feel like with AI people can have less knowledge,” Zhang said. “The new skill is looking for knowledge … I feel like AI is really changing the skill of how to get something done right faster.”

“There are different kinds of skills, and there’s different kinds of knowledge,” said Skiena. “If we’re thinking about knowledge as opposed to artifact-producing, do I believe that I can get the AI to write me a document quickly that’s producing an artifact? Does it help me understand that artifact? The answer is no. It’s important that people understand what is being produced, not just the model that’s producing it.”

Zhang agreed, adding, “If human beings don’t have the knowledge to tell whether the artifacts are true or not, it’s going to be a total disaster.”

“If we’re in a world where the model knows everything, I don’t have to know anything,” added Skiena. “That would be unfortunate. The trick is going to be making sure we put these things at the right place in the educational pipeline.”

Robert Emproto
SBU News