�鶹��Ѱ�����

�鶹��Ѱ�����Engineering strives for an intentional approach to incorporating AI in its research and teaching mission

When it comes to engaging with artificial intelligence in education, �鶹��Ѱ�����Engineering is “unique,” said Sriram Sankaranarayanan, associate dean and professor of computer science. His Office of Digital Education places “a lot of focus” on how his colleagues incorporate AI into curricula.

“If the future requires students to be good at using AI and also good engineers,” he said, “then we train students to be strong in both.”

Across the college, PhD students use AI to build interactive diagrams from static textbooks, researchers study its biases and costs, and students are enrolling in a new professional master’s program focused on AI. Faculty have created space to help one another learn about AI, while others encourage students to use it to refine papers and iterate ideas.

Illustration by Hanna Nordwall

The why of AI in the classroom

Diane Sieber, an associate professor in the Herbst Program for Engineering, Ethics and Society, runs the Generative Futures Lab, an informal space where faculty, staff and students can explore AI. She developed a framework to help instructors decide what level of AI use is appropriate for different assignments.

“We even made icons for syllabi, ranging from ‘full AI use encouraged’ to ‘AI use restricted,’ with explanations for each,” Sieber said.

Alvaro Velasquez, assistant professor of computer science, has already seen the benefits of AI in graduate-student writing. “They write the paper, and by iterating with AI, they refine it to a point where it’s actually very, very well written,” he said.

For students with AI anxiety, Sankaranarayanan asks they “treat this as a challenge.”

“Try to gain expertise beyond what AI can replace,” he says. “Think broadly about opportunities and the skills you need to thrive, so you don’t feel victimized by a changing technological landscape.”

Making STEM accessible through AI

Many faculty see AI as a tool for broadening access to STEM education. Velasquez believes AI could eventually help close gender and socioeconomic gaps by functioning as a personalized tutor.

To that end, ATLAS Institute Assistant Professor Ryo Suzuki is developing tools such as Augmented Physics, which uses AI alongside augmented and virtual reality to turn static textbook diagrams into interactive simulations.

“Now, imagine if AI could not only respond in text but also generate animated explanations in real time, going beyond text to make learning engaging, visual and playful,” Suzuki said.

Thomas Breideband, associate director of the NSF National AI Institute for Student- AI Teaming, works with an interdisciplinary community of researchers to study how AI can support collaboration in learning environments — particularly in middle and high school, where STEM participation gaps often emerge.

“Collaboration is a critical 21st-century skill, but it is very rarely explicitly practiced,” Breideband said. He and his colleagues are currently working with schools to integrate AI-enhanced curriculum units in classrooms.

AI in the lab

Researchers also interrogate AI. In Associate Professor Daniel Acuña’s lab, his team uses AI techniques to study how large language models absorb “prestige bias” from language used to describe scientific research.

“Instead of saying, ‘scientists found X,’ journalists say, ‘Harvard scientists found X,’” Acuña said. Because LLMs learn from that same data, Acuña hopes his work can help highlight — and eventually correct — those patterns.

Velasquez is tackling another challenge: AI’s immense computational cost. His research on neurosymbolic AI seeks to end “gatekeeping” and "monopolization" in AI.

“When it comes to natural language generation, it comes at a tremendous cost,” Velasquez said. “The neurosymbolic AI community has seen cases where we can use 100,000 to a million times less data.”

New AI degree program

The newly launched Professional Master’s in Artificial Intelligence merges technical skills with ethical considerations.

“If one of our graduates is hired by a company like Walmart, they should be able to analyze consumer data to decide what products to buy and how to manage inventory,” Sankaranarayanan said. “But they should also be able to say, ‘This approach preserves privacy, complies with regulations and is ethically sound.’”

The online degree option through Coursera is up and running, while the on-campus version begins in fall 2026.

Educators agree that they must be intentional in their approach to teaching AI.

“Technology has a very mixed record,” said Sankaranarayanan. “People fear losing jobs, and that is a valid fear, but it doesn’t have to go that way. That’s where education becomes extremely important.”