A closeup photo of the Midjourney home page taken with a special camera lens. Midjourney is a form of generative AI for making images, and has had a starring role in the surge of interest in AI the past two years. Rhode Island College recently formalized an AI degree program to help students master the skills needed to build programs like Midjourney, which rely on large language models, which are massive data sets used to generate patterns. (Alexander Castro/Rhode Island Current)
This article is part of an ongoing series on AI in Rhode Island.
A favorite puzzle of philosophers is the trolley problem. You can pull a lever and divert a runaway trolley just in time to avoid killing five people — but one person would die on the new track. Do you pull the lever?
On a sunny September morning in a third-floor classroom on the Providence campus of Rhode Island College (RIC), small groups of students debated a tweaked trolley problem in which a robot is at the switch. They’ve all read sci-fi author Isaac Asimov’s allegorical “three laws of robotics,” which deemed it forbidden for robots to hurt humans. What’s a bot to do, when a bloody outcome seems unavoidable?
Sonya Cheteyan thought the dilemma wouldn’t be any easier for a machine to process.
“If it’s impossible for a human, it’s going to be very hard for an AI to figure that out, because AI is just an extension of our intelligence,” said Cheteyan, a RIC junior double majoring in computer science and AI and a lifelong Providence resident.
“It’s like a combination of everyone’s intelligence packed into one entity to make decisions for us. And it’s hard, because if we don’t have an answer to that problem, the AI won’t.”
Cheteyan encountered the trolley problem in the course “Application and Impact of AI,” taught by Associate Professor Leonardo Pinheiro. It’s one of two courses integrating ethical detective work in RIC’s new AI curriculum, which launched in September.
The bachelor’s of science program is the first such AI-focused undergraduate degree at a Rhode Island state school. It’s meant to prepare students for a workforce in which AI is no longer novel, said Tim Henry, the associate professor who leads the new AI program. That means educating students on not only how to build artificial intelligence, but how to do it responsibly.
“We’re going to be educating Rhode Island students, helping them get prepared for the workforce,” Henry said. “Businesses already are becoming comfortable in using AI. The hard thing is understanding how to use it well, and being able to know what the risks are in how you’re using it.”
What students learn in the AI program overlaps and complements the existing computer science major. “[They’re] supplemental in a way, because computer science skills are just as important as knowing how AI models work,” Cheteyan said.
Greater AI literacy is already on the agenda in the K-12 sphere, as seen in bills like the LIFT AI Act, co-sponsored by Rhode Island’s Democratic U.S. Rep. Gabe Amo and New Jersey’s Republican Rep. Tom Kean Jr. Classes like Pinheiro’s at RIC, meanwhile, help contextualize artificial intelligence for those who will be entering the workforce much sooner. Cheteyan said she didn’t expect to enjoy the course as much as she has.
“He really unfolded everything for me,” she said of Pinheiro. “‘I’m like, ‘Oh, AI is more than just ChatGPT.’ It’s used in data science. It’s used for facial recognition. It’s been used before it was a popular word…I was surprised how far back it went.”
Let a=4
The AI program is joined by two AI-focused minors and synchronizes with the school’s Institute for Cybersecurity & Emerging Technologies, which opened last November and will benefit from a ballot question approved by 59.7% of voters in the Nov. 5 general election.Voters said yes to a $160.5 million bond issue for capital improvements at RIC and the University of Rhode Island. RIC would use its $73 million share to renovate and modernize Whipple Hall into a dedicated space for the institute, with updated computer labs, data centers, IT infrastructure, classrooms, and cyber ranges where students could model threats in real time.
The updated facilities would differ vastly from where Henry typed his first code in 1976: the inside of his high school’s janitorial closet, where he and three fellow students worked on a Teletype terminal connected to the University of Pennsylvania.
“We got to write our first computer program,” Henry said in a recent interview at his Alger Hall office on the RIC campus. “It was like, ‘Let a equal four, let b equal five, let c equal a plus b, print c.’”
That was the initial site of Henry’s fascination, and his parents gave him a four-function calculator for Christmas. After service in the Coast Guard and a gymnastics career that included two All-American status, today the professor is in high demand for his sagacity with code.
“Dr. Henry’s record speaks for itself,” said Suzanne Mello-Stark, chair of the Computer Science and Information Systems department at RIC and a former Ph.D. student of Henry’s at URI.
That’s why RIC recruited Henry to lead the AI program. Henry also serves on Gov. Dan McKee’s Task Force on Artificial Intelligence, which was created by executive order in February and met for the first time in July. The task force is chaired by former U.S. Rep. Jim Langevin, who serves as distinguished chair of RIC’s cybersecurity institute. He, too, wanted Henry on board for the state’s first executive-level initiative about AI, which will eventually produce a report with findings and recommendations for how state government should best proceed with AI — a set of technologies that has been heralded as both apocalyptic and messianic.
As Cheteyan sees it, chatter about AI can be excessively negative. “There’s some doom, for example, in journalism,” she said, where there’s fear that AI could replace or scrub out certain jobs.
But Cheteyan believes AI can be leveraged as a tool for good rather than ruin. “The goal is to be educated on how to use it, and how to take advantage of AI, instead of sitting back and being like, ‘Oh, now I lost my job.’ No, we just made your job easier,” she said.
Computers can be artists, too
Generative AI has made more people realize, though not always welcome, the ability of AI to make songs, images and stories. A little under two years after ChatGPT emerged as a sophisticated successor to the chatbots of yore, the hype over generative AI — which creates content — has not evaporated. But even a well-trained robot can’t solve everything. What if you ask ChatGPT to “solve the trolley problem”?
“It’s unlikely a robot would fully ‘solve’ the trolley problem, as it’s fundamentally a question of moral philosophy rather than pure logic,” ChatGPT wrote back when prompted.
Henry said the creative opportunities help generate much of the interest in AI.
“If you have AI predict the sales of your organization in two months — OK, great,” Henry said. “That’s really interesting to a small, narrow group of people. But if AI can converse with you, or generate some text for you, that’s useful to a lot more people. If it can generate an image to you, it’s useful to a lot more people.”
Coding, Cheteyan said, is “actually a type of creativity. I know a lot of people say it’s math, but you have to think, you have to come up with good solutions to problems, and you need a creative mind for that.”
Two projects she’s coded are a chatbot for popular messaging app Discord, and a model that analyzes Netflix data to predict a user’s watch time in a given day. Cheteyan works mostly in the programming language Python, which is “necessary” for many large language models AI uses, and it prioritizes clean and readable code.
Per one of the 20 koan-like guidelines known as the Zen of Python: “In the face of ambiguity, refuse the temptation to guess.”
A job to make other jobs easier
But there is still guesswork involved in what titles or positions AI majors will hold when they graduate, Henry said.
“It’s a very good question, and everybody’s trying to figure that out,” Henry said. “I can’t throw out necessarily titles, but sometimes it’s ‘computer vision engineer’ or a ‘robotics engineer’ or a ‘natural language processing engineer.’ Those are very specific areas, but it’s generally going to be somebody who does machine learning as a software engineer. So software engineering is still the general category.”
Data science is one field where AI is making a noticeable impact, Henry said, because “data analysis and preparation, exploratory data analysis…those are kind of the first part of any good AI model, being able to do those well.”
He also noted AI’s utility in cybersecurity, which makes it a natural fit for RIC’s institute: “AI tools are really good at being able to understand and learn normal network traffic patterns, for example, or behaviors that are typical of malicious software.”
Cheteyan is likewise unsure of her exact future, or what job she’ll one day hold, but said she hopes to build “helpful tools” like chatbots or leverage data science to make predictions for companies. She offered the example of sentiment analysis, which scours massive amounts of text to discourse online about certain products, people or businesses.
Cautious optimism surrounds generative AI, highlighted recently by Federal Reserve Governor Lisa D. Cook, who called it a “general-purpose technology,” that could increase productivity, lower inflation and be a boon for the job market overall, albeit one that would take time.
A total of 386 students are now taking classes under the AI umbrella through the institute. That number includes a mix of majors, double majors, and minors, said Lindsay Russell, a spokesperson for the Institute for Cybersecurity & Emerging Technologies..
An exact number of majors wasn’t yet available, and Henry said the school is still enrolling majors. “Most of the students I know of right now are people who were here as sophomores and enrolled as an AI major their junior year,” he added.
In an upcoming semester, Henry will teach a course on ethics in AI technologies.
“The bigger problem is understanding how the software can be used outside of what you plan it to be used for,” Henry said. “That’s our ethical responsibility, to look at the data and how we’re gathering data to train it.”
GET THE MORNING HEADLINES.