A home for responsible optimists

When Deblina Sarkar checked her email one afternoon this past February, she saw a heart-wrenching email from a father. He had found his way to the website of her Nano-Cybernetic Biotrek research group at the MIT Media Lab and written to tell her about his 10-year-old daughter. The girl had diffuse intrinsic pontine glioma (DIPG), an aggressive form of brain cancer resistant to standard treatments. He had come across Sarkar’s research on nanoscale electronic devices designed to move through the bloodstream, reach diseased regions of the brain, and deliver precisely targeted electrical stimulation, and he wanted to know if there was any hope.

Messages like this one land in her inbox every other week or so. It’s “the main motivation behind our work,” says Sarkar, the AT&T Career Development Associate Professor of Media Arts and Sciences and a member of the MIT Center for Neurobiological Engineering.

Sarkar started her career thinking about nanoelectronics. A physicist and electrical engineer by training, she spent years making ever-smaller transistors, chasing increasingly low power consumption, until she recognized that the most energy-efficient computer on Earth was actually in the human skull. The brain runs on roughly as much power as a dim lightbulb, and Sarkar says modern computing systems consume about a million times that amount to do comparable work. But the lack of treatments for brain diseases got her interested in building tools to understand—and heal—the brain. She began to think about how nanoscale electronics, fused into biological structures, might do things medicine had never managed.

The result, developed over more than seven years, is a platform she calls circulatronics: nanoscale electronic devices capable of navigating the body’s fluid systems, identifying diseased tissue, and providing wireless therapeutic stimulation—no surgery, holes in the skull, or $100,000 procedure required. In collaboration with the Mayo Clinic, she’s shown in mice that precise electrical stimulation can halt tumor growth in cases where standard treatments have failed. The technology may not arrive in time for the girl whose father wrote to her. But Sarkar believes clinical trials to test the platform could begin within three years. 

“The main message,” she says, “is that we want to make life-saving technologies accessible to all, irrespective of their socioeconomic status.”

Sarkar’s work comes at a pivotal moment for the Media Lab—and for technology itself. AI is reshaping computer science, policy, psychology, medicine, education, and art all at once. It touches the body in Sarkar’s work, the mind in others’, and it raises questions of access and power elsewhere. The Media Lab, now starting its fifth decade, gives those topics room to intersect and the people pursuing them permission to follow the work wherever it leads.

A new era of leadership

This October, the Media Lab will bring together its global community of researchers, faculty, students, and alumni to celebrate its history and look to the future. The event caps off the lab’s 40th year, which began just as Dava Newman, SM ’89, PhD ’92, the Apollo Program Professor of Aero-Astro and former deputy administrator at NASA, was wrapping up her tenure as director. 

Upon taking the helm in 2021—after a worldwide search for a successor to Joi Ito, who stepped down in 2019—Newman helped the MediaLab regroupand prepare for its next four decades. As part of that effort, she split the lab’s directorship into two complementary positions: an executive director to oversee fundraising and communications, and a faculty director responsible for research. Working together, the two would shepherd the lab’s vision and give its researchers the support to realize it. 

Tod Machover, the Muriel R. Cooper Professor of Music and Media and director of the Opera of the Future research group, stepped into the role of faculty director in July 2025. It was a logical fit: Machover, a composer who also builds electronic instruments (including a hypercello for Yo-Yo Ma), has been on the Media Lab faculty since the mid-1980s. 

“If technology is changing, we should wrestle with it and figure out what good we can do for humanity.”

Then, in September, Jessica Rosenworcel became the lab’s executive director after stepping down as chair of the Federal Communications Commission. At the FCC, she was accustomed to seeing technology deployed in ways that were already shaping public life for better and worse. For example, the ability to move data across networks underpins much of the modern economy, but she also saw that technology being used to generate robocalls, in one case even turning a clone of President Biden’s voice into a tool for election interference. At the Media Lab, she’d get a chance to help steer new technologies in positive directions. 

“This is a place invested in the idea that if technology is changing, we should wrestle with it and figure out what good we can do for humanity,” she says.

As Rosenworcel settled in, she and Machover found that their different strengths complemented each other. “I’ve really not seen anybody who can get to the core of a problem and synthesize complex situations as Jessica can,” Machover says. Meanwhile, she sees in Machover the intelligence of an artist who won’t let something go until it is exactly as it should be. “The perfection of small things can make such a difference in a piece of art,” she says. “There’s incredible focus there.”

Two months after Rosenworcel arrived, the Media Lab/Media Arts and Sciences Visiting Committee, an outside group of academics and businesspeople including alumni and members of the MIT Corporation, came to evaluate the lab. “It forced me to learn this place really fast,” Rosenworcel says, “and get us to really think about the story we want to tell going forward.” 

For her, the pace of the transition fit the pace of the moment. “This place [has] thought largely about technology and deploying it at scale,” she says. “But now we also have to think about velocity, because change is coming faster.” 

Being digital and being human

The Media Lab opened in 1985, the year after Apple launched the Macintosh and just as MIDI, the evolving standard that allows digital instruments and computers to communicate, was becoming part of the infrastructure of electronic music. The lab’s founders—Nicholas Negroponte ’66, MArch ’66, a young professor of computer graphics with a contrarian streak, and Jerome Wiesner, a former MIT president—believed that computers would become creative and communicative tools, changing how people made art, shared information, learned, and understood the world.

“What’s the future of being a human being? That’s the question everybody here is asking.”

That founding thesis proved durable. Again and again, the Media Lab moved early into emerging technologies—when there was still room to shape what they could become, who could use them, and whose interests they would serve. In the 1960s, Seymour Papert, one of the lab’s founding intellectual spirits, helped develop Logo, an educational programming language that let children learn mathematics by writing commands for a small on-screen turtle or a turtle-like robot. His student Mitch Resnick, SM ’88, PhD ’92, carried that mission into Scratch, now the most widely used tool in the world for teaching children to code. In 1998, Neil Gershenfeld launched “How to Make Almost Anything,” a course that catalyzed the global fab lab movement, spreading digital fabrication tools from Somerville to the Norwegian Arctic. The technology behind the Kindle’s E Ink display was spun out of the lab in 1997. In each case, the animating idea was consistent: Powerful tools should reach everyone.

Today, the lab is facing a new set of urgent, constantly evolving questions. As AI reshapes the world, it’s asking what role it should play in keeping the future open, creative, and humane. “Nothing compares to what’s happened in these last few years,” Machover says. Generative AI has become both a growing research focus and a daily presence for millions, with more than 61% of US adults 18 to 64 reportedly using it.

“The way AI is developing in so many instances actually shuts people out,” he says. “The most powerful systems aren’t systems we can build ourselves. They’re not really designed to allow us to use what makes us individuals. What’s the future of being a human being? That’s the question everybody here is asking.”

The measure of flourishing

Pat Pataranutaporn, SM ’20, PhD ’24, grew up in Thailand playing with toys like Lego Mindstorms, the robotics kit that grew out of Seymour Papert’s ideas about learning by making. In high school, the lab popped onto his radar when he watched a TED Talk by Pattie Maes, the Germeshausen Professor of Media Arts and Sciences at the Media Lab, who had helped develop wearable technology that could offer people relevant information about things they encounter in the physical world. Her talk was one of the most watched ever; what stayed with him, he says, was how Maes “really embodied the idea that we can bring technology, design, engineering, art, and humanities together to think about the vision for the future.” 

After earning his undergrad degree at Arizona State University, he joined Maes’s Fluid Interfaces group as a graduate student, completing a master’s and then a PhD. Today, he is one of the lab’s newest faculty hires, running a group called Cyborg Psychology.

Pat Pataranutaporn, SM ’20, PhD ’24, is studying what happens to the human psyche when the boundaries between people and technology blur.

KATHERINE TAYLOR/TEDX BOSTON VIA MIT MEDIA LAB

While Pataranutaporn sees the value of AI as a productivity tool or a coding assistant, he wants to understand what happens to the human psyche as the boundaries between people and technology blur. What does it do to a person’s sense of agency when a machine can answer any question? What becomes of curiosity when search is instant? What happens to emotional life when algorithmic companionship is available day or night? 

Current AI benchmarks, he believes, often measure the wrong things. “A lot of people today are focusing on sort of a narrow area, like productivity,” he says. “But I think overall, what we need as human beings is flourishing.” A model might ace a calculus problem while leaving the student who used it less capable of solving the next one. A chatbot might pass every safety review and still steer someone in crisis away from help. 

So alongside Maes, he launched the Advancing Humans with AI (AHA) research initiative to develop benchmarks that measure AI’s effects on people. His group studies whether an AI system’s responses support human outcomes like agency, curiosity, mental health, physical health, and sense of purpose. The research has already shaped policy, including California legislation regulating companion chatbots.

Early in his PhD studies, before ChatGPT had become mainstream and when the technology was still largely associated in the public mind with deepfakes, Pataranutaporn co-designed and taught one of the first MIT courses on generative AI. The class explored not only its technical aspects but also its creative and ethical dimensions. 

At the Media Lab, he also works with dancers and choreographers to build virtual avatars that perform Thai classical dance, using technology often associated with displacing human work to preserve a living cultural tradition. And he has explored what he calls “deepfakes for good,” asking whether generative AI tools that often propagate misinformation could be turned toward creative and constructive ends—for example, animating digital companions for children undergoing medical procedures, answering their questions and providing emotional support. 

“How we shape technology really matters,” he says. “We should not let the technology take the wheel.”

Data for justice

Media Lab graduate student Raechel Walker, SM ’23, is rising to that challenge by studying what happens when the people building AI systems don’t consider the communities those systems affect.

In her undergraduate data science classes at UC San Diego, Walker noticed she was usually the only woman, often the only Black student, and almost always the only person whose parents hadn’t worked in STEM. “I realized that there was just a huge lack of diversity within the space of data science and AI,” she says, “and I also saw that this was leading to a lot of AI harm.” 

Her coursework and research also led her to observe that some AI systems aren’t good at recognizing darker-­skinned faces and work less reliably for people who speak English as a second language. Though such patterns predate AI—Kodak’s film stocks, for example, were calibrated for lighter skin for decades—it’s yanked them back into the foreground.

Walker, now in the final stretch of her PhD program in the Personal Robots group of Cynthia Breazeal, SM ’93, ScD ’00, works on what she calls liberatory computing: teaching young people from underrepresented communities to use data, AI, and computing as tools for civic action and to mitigate systemic oppression. The curricula she’s developed combine rigorous data science with lessons on embedded racism; they give students ways to analyze the systems around them and help them see that they can challenge the power dynamics in the way AI is used.

Raechel Walker, SM ’23, says AI education should teach students to ask questions like who built a system, whose data it uses, and what assumptions it carries.

JAKE BELCHER/MIT MEDIA LAB

In her Data Activism Program, low-income high school students collaborate with social justice organizations on projects with real stakes. For one, Walker had students partnerwith the Algorithmic Justice League (founded by Media Lab alum Joy Buolamwini, SM ’17, PhD ’22)to study mass surveillance in their schools. The students had noticed the proliferation of security cameras and tracking software but couldn’t say precisely why it made them uncomfortable. After studying the use of surveillance in slavery and its role in what’s known as the school-to-prison pipeline, students conducted more than 30 interviews with teachers, principals, and peers, analyzed national data sets, built data visualizations, and wrote a research report that the Algorithmic Justice League subsequently used to advocate for changes in how student data is collected and retained.

Walker’s latest research focuses on reflexivity, the principle that people deploying an AI system in a community-facing context need to first interrogate the assumptions embedded in their own background. She argues that what constitutes responsible AI cannot be captured by a fairness score alone: “People should be more aware of how history is contributing to some of these harms that AI is amplifying.” 

In Walker’s view, AI education should teach students to ask who built a system, whose data it uses, what assumptions it carries, what harms it may amplify, and what to do with those answers. Even if students don’t have a technical background, “they can still contribute a lot,” she says.

That argument is gaining traction within the lab. Last fall, Walker took part in an AHA workshop Pataranutaporn helped organize that brought together researchers from institutions including MIT, Stanford, and Oxford, along with leaders from OpenAI, Microsoft, Google, and other companies, to develop AI benchmarks centered on human flourishing. Ultimately, the group concluded that such standards could not be defined by such a small group of people. 

Keeping the future open

The Media Lab, Machover believes, is unusual in having sustained its energy for 40 years. He notes that many legendary experimental spaces—the Bauhaus, Bell Labs, Xerox PARC—had a burst of creativity that ran its course within a few decades. What has kept the lab vibrant is a disposition more than any particular technology: the impulse to find a powerful idea, demonstrate it compellingly, release it to the world, and move on. 

He says that while most companies or academic labs try to hold onto a good idea to develop it and “make a lot of money,” the Media Lab tends to favor an open-source approach. Scratch is free. Fab labs spread because Gershenfeld wanted them to. The lab’s ethos is to “find something beautiful and fantastic,” he says, and “get to the point where its excitement is palpable and then kind of give it away.” 

That mindset shapes the lab’s makeup. Sarkar arrived as a nanoelectronics specialist and rebuilt herself into a neuroscientist and then a cancer researcher, because the problems she pursued kept forcing her across disciplinary lines. “We are a place of misfits,” she says. “I was a fit in the Media Lab because I was a misfit.” 

Her circulatronics technology begins with electronic devices so small they can move through the body’s smallest blood vessels. The devices are fabricated like tiny chips and then chemically bonded to living cells. For example, Sarkar’s team has used monocytes, immune cells that naturally travel toward inflammation, to help hide the electronics from the immune system, carry them through the bloodstream, and allow them to cross the blood-brain barrier without damaging the barrier itself.

Once the cell-electronics hybrids reach an inflamed region of the brain, they implant themselves there. In the monocyte work, an external transmitter then sent near­infrared light through tissue to power the devices wirelessly, allowing them to electrically stimulate nearby cells without a surgically implanted battery or electrode. This neuro­modulation—as the process of stimulating neurons is called—is considered a promising method for treating brain cancer, Alzheimer’s, chronic pain, blindness, and multiple sclerosis. In mice, the hybrids reached precisely targeted regions of brain inflammation, responded only when wirelessly activated, and did not appear to impair movement, cognition, blood chemistry, or functioning of major organs. 

Existing brain implants can deliver such stimulation but require surgery and specialized equipment, and they’re expensive. Circulatronics is an attempt to give that same type of therapy—as well as therapies using different cell types that biochemically target other diseases—a new route into the body. 

The work also reflects Sarkar’s broader interest in ultra-low-power hardware for artificial intelligence. Over time, she wants to develop AI-enabled systems small and efficient enough to sense what is happening in tissues, process that information, and respond inside the body. In her work, AI is not a chatbot or a screen interface but a set of capabilities that could eventually help therapies act in places drugs and scalpels cannot easily reach. Indeed, Sarkar says the technology, which could also be used for wireless and surgery-free pacemakers, could reach every nook and cranny of the body.

While she sees the Media Lab as a haven for misfits, Machover describes incoming master’s students as people who often arrive with “two or three or four different qualities,” not yet sure how their interests connect. “We look at our master’s program as a place for students to find out who they really are and what they really care about,” he says.

“Everyone here wants to be a responsible optimist,” Rosenworcel says. “There are a lot of reasons right now to be pessimistic about our world and its politics. It can feel kind of dark, like something that strips us of our agency, constrains our privacy, limits our options. And I think instead of cowering in the face of all of those troubles, we’re like: No. We’re going to take these forces, we’re going to wrestle with them, we’re going to try to come up with something that’s good.”

In her own lab, Sarkar is still thinking about the father of that 10-year-old, but she’s also anticipating the next email. She’s thinking about the roughly 3 billion people who live with neurological diseases that are often difficult to treat, about the devices that might one day find their way through the bloodstream to diseased regions of the brain that surgery cannot easily reach.

“I envision a future,” she says, “where human beings will not be slaves of their biological limitations.”