Artificial Intelligence

Today’s revolution in artificial intelligence didn’t come out of nowhere. Decades of research provided the theoretical and technical backbone for the massive scaling of generative AI models that are changing how we work and learn today. As computing power, data, and technology continue to evolve, SEAS researchers are focused on building the next generation of AI that is more capable, trustworthy, and efficient — AI that has the capacity to advance and enable scientific research itself.

Our researchers are working across disciplines to build the foundations and infrastructure for this next leap forward in capability. We are building AI that integrates with sensors, wearable devices, and equipment to plan, execute, and analyze tasks in real time. We are harnessing AI to pioneer advances in optics and materials design, synthetic biology, drug discovery and biomechanics, energy system modeling and more.

At SEAS, we aren’t following an AI trend — we are building the foundation for what comes next.

The possibilities enabled by AI will come more quickly than we can imagine. At SEAS, we are pushing the frontiers of AI to advance science and engineering, transform education, and bring positive impact to the world.

David C. Parkes

John A. Paulson Dean

AI For Science And Engineering

At SEAS, AI is increasingly woven into the heart of scientific discovery. Researchers are using AI to do new kinds of science and to accelerate traditional experimental and theoretical work. Our researchers are using AI tools to make breakthroughs in materials design, bioengineering, cellular therapeutics, environmental science, and more. Here, we are training an “AI‑first” generation of scientists and engineers who treat intelligent tools as standard instruments of discovery.

visualization with arrows and spheres of atomic charge dynamics in barium titanate

AI Model Simulates Million‑Atom Electric Response To Speed Next‑Gen Device Design

Machine‑learning framework predicts, with quantum‑level accuracy, how materials respond to electric fields at the million‑atom scale

Algorithms Empower Metalens Design

New approach paves the way for larger, more complex metalenses

Machine Learning Helps Trace Climate Change’s Role In Growing Wildfire Smoke Crisis

Using machine learning with decades of fire, weather, and air-quality data, researchers quantified how warming intensified U.S. wildfires and smoke pollution

artist's representation of disordered proteins

AI Reveals Hidden Order In Disordered Proteins

Physics-based machine learning algorithm applied to biomolecule design

Optimizing How Cells Self-Organize

Computational framework could lead to organ design, cellular programming

A glowing red waveform fluctuates across a dark, grid-lined computer screen, resembling a data or signal trace overlaid with numerical markers.

Study Introduces Innovative Method to Decode Complex Neural Data

A new machine learning framework helps disentangle information patterns in the brain

How Is AI Changing The Field Of Synthetic Biology?

Mo Khalil, the Hok Lam and Kathleen Kam Wong Professor of Bioengineering, explains how machine learning is influencing drug development and the study of biological molecules.

The Future is Physical

At SEAS, researchers are building a future where AI-powered devices work alongside human teammates in labs, factories, clinics, remote research bases and more. Our researchers are connecting advanced learning models with real‑world sensors, robots, and wearable devices to create AI systems that can perceive, decide, and act in dynamic environments. By developing methods that let machines learn from human motion, interpret rich visual and sensory data, and coordinate across teams in uncertain settings, SEAS is advancing physical AI — making it safer, more adaptive, and more reliable to deploy in real-world environments.

White industrial robot arm with a gripper reaches forward against a blue background filled with glowing network lines and nodes, suggesting AI‑driven automation.

A New Kind Of AI Model Gives Robots A Visual Imagination

Yilun Du and his team introduce a breakthrough system that could redefine how robots learn to act in the physical world

Building 3D Scenes From 2D Images In An Instant

New algorithm speeds up computer’s reconstruction capabilities

Close-up of a person’s hands in blue and purple lighting as they tap the screen of a smartwatch on their wrist.

Generative AI At The Edge: Challenges And Opportunities

Professor Vijay Reddi explains what it takes to deploy AI agents in physical settings and how to make them work safely at scale

A Wearable Robot That Learns

New control algorithm personalizes user experience for stroke, ALS patients

illustration of a fleet of autonomous trucks

Do You Trust Me? A Framework For Making Networks of Robots and Vehicles Safer

Paper describes how 'cy-trust' can enable resilient systems

AI-Enabled E-Nose Sniffs Out Harmful Molecules

Portable sensor can be used to manage hazardous waste, monitor air quality, identify spoiled food, and diagnose diseases

AI Robotics & Wearables: The Next Frontier in Care

Patrick Slade, an Assistant Professor of Bioengineering, provides insights into the rapidly advancing field of AI-enabled robotics, showcasing how assistive technology is being developed to support rehabilitation.

AI Innovation And Entrepreneurship

SEAS alumni are building companies that turn cutting‑edge research into real-world impact. Drawing on strong foundations in computer science, applied mathematics, and engineering, our alumni are developing AI to make complex work more efficient and help professionals in fields from healthcare to infrastructure and industry. Many of these ventures grew directly out of projects that began at SEAS, where students learn to pair technical depth with entrepreneurial thinking.

Harvard SEAS HBS alum Bryce Strauss with start-up co-founders Cameron McCord and Jason Hoch sitting at a laptop

Building Better Tools for Complex Engineering: MS/MBA Alum Co-Founds Billion-Dollar Start-Up

Nominal offers AI-driven data analysis tools for aerospace, defense and more

Kyna Fong Builds Elation Health To Streamline Primary Care For 20 Million Patients

A.B./S.M. alum is reshaping how doctors manage patient relationships and deliver care

Next-Gen AI Systems: Stronger, Safer And More Aligned

The next generation of AI systems will need to be faster and more powerful than current models, as well as safer, more transparent, and aligned with human values. At SEAS, researchers are attacking these challenges from multiple directions: using neuroscience to inspire new model architectures, developing techniques that expose and correct hidden behaviors and biases, and designing new AI systems that support, rather than replace, human learning, judgment and creativity.

How Brain Waves Are Inspiring A New Kind of AI Vision Model

Researchers draw inspiration from neuroscience to design neural networks that integrate spatial information over time and solve complex image segmentation tasks more efficiently

Person typing on a laptop with a translucent purple overlay showing a “Chat AI” interface and floating speech bubbles, symbolizing interaction with an AI chatbot.

How Do AI Chatbots See Us?

In this video, Professors Fernanda Viégas and Martin Wattenberg discuss their dashboard which allows end-users to see — and control — how a chatbot perceives them

Close-up of a computer screen showing source code with an “AI Actions” context menu open, highlighting options such as “Explain Code,” “Suggest Refactoring,” and “Find Problems,” illustrating AI-assisted programming tools

How Confessions Can Keep Language Models Honest

Professor Boaz Barak and collaborators at OpenAI share an early, proof-of-concept method that trains models to report when they break instructions or take unintended shortcuts.

Group of suited delegates sit in a row at a conference table wearing headsets, reading documents, and speaking into microphones during a formal meeting

Designing AI for Humanitarian Frontline Negotiations

Researchers outline design principles for deploying AI in high‑stakes fields where context, ethics, and trust are as important as outcomes

Retro 8‑bit platform game screen showing a small character climbing a tall green ladder between brick platforms. A flaming red monster sits on the middle platform, while a blue swinging object hangs on a rope to the right. Additional ladders and a small box appear below, all framed by a colorful patterned border with the text “© SHAMIS” at the bottom.

How To Optimize The Training Of Large Language Models? Use Resets

Assistant Professor Kianté Brantley studies how resetting LLMs can make reinforcement learning more efficient

Alt text: Person in a dark suit holding a digital tablet with a stylus, overlaid with glowing blue icons representing law and technology, including a central scale of justice, a gavel, and abstract AI and data symbols, suggesting the intersection of artificial intelligence and legal decision‑making

SEAS Researchers Expose Hidden “Alignment Discretion” Shaping AI Behavior

Research demonstrates how human judgment steers AI safety training and calls to make alignment more transparent and accountable

AI + Societal Challenges

Learn how SEAS researchers are using AI to tackle some of society's biggest challenges.

Can AI Strengthen Democracy And Improve Collective Decision-Making?

A Q&A with Professor Ariel Procaccia

Is AI the Problem And The Solution To The Aging Power Grid?

A Q&A with Professor Le Xie

In the Classroom and Beyond

In SEAS classrooms, maker spaces and lecture halls, students are learning how to use AI systems critically — probing their limits, checking their work, and understanding the ethical stakes of deploying them in the real world. New and redesigned courses invite students to experiment with AI for programming, data analysis, and design, while also tackling questions of bias, safety, and alignment. In collaboration with centers and institutes from across the University, including the Harvard Grid, the iLab and more, students gain opportunities to build novel AI applications and study their impact — preparing them to be thoughtful and innovative leaders in the field.

Mission in Action