SERC Projects, 2026-2027
| Group Leader | DLCI | Project Description | Meeting Times |
| Dr. Natalie Collina | CSAIL | Learning to Play Well with Others: How can you make algorithms that are good at games? What does it mean for an algorithm to be good at a game? What do we gain–and what do we lose–when we play games with algorithms? Throughout this course we will play games, both with and without algorithms. We will learn about key technical tools–such as control theory, reinforcement learning, no-regret learning, and neural networks–which have shaped, currently define, or may define the future of algorithmic game-playing. We will explore what makes games a unique space for algorithm development, both from technical and social perspectives. For your final project, you and your group will build an algorithm that plays a game. With my guidance, you will choose both a game and a metric for evaluating your algorithm. This metric may be its win rate, how enjoyable it is to play against, how well it teaches a human player the rules, or something entirely different. Even “win rate”can mean many things. Should your algorithm perform well in the worst case, or on average? Should it succeed against humans, or against other algorithms of a particular structure? After deciding what success means, you then design and implement your algorithm for the game, and analyze it theoretically and experimentally. | Mondays, 2:00 – 3:00 PM |
| Dr. Sarah Gillet Schlegel | MIT Media Lab | Technology for Connection and Culture: Strong relationships are one of the most reliable predictors of good mental and physical health, and even of how long we live. Yet much of today’s technology is designed to be used alone: scrolling a feed, or talking to a chatbot that agrees with everything you say. A popular story says this technology is taking away from our relationships and making us lonelier. In this SERC Scholar group, we will see that the evidence is messier, and more interesting, than that story. It’s also not universal: even basic questions, like whether loneliness is really rising or what counts as healthy connection, look different across cultures. We take an interdisciplinary look at technology and human connection as an open question. We start by reading current research on how commercial technology really affects the way people relate to each other, engaging with loneliness and isolation, the effects of phones on interactions, and what happens when the “other party” is an AI. One theme we’ll keep returning to: what actually makes people feel connected is being genuinely responded to, the real back-and-forth that reacts to you, which is exactly what feeds and agreeable chatbots fail to provide. We then turn to a harder question: can technology support human connection rather than replace it? Here we’ll distinguish technology meant to be your relationship from technology meant to help people connect with each other, acting as a mediator between people rather than a substitute for them. In the second half of the group, you’ll bring this knowledge into your own project. You’ll choose a technology, such as the social robot Jibo, a chatbot, or a phone app, and explore how it could be reshaped to act as a mediator that connects people with each other. | Mondays, 4:00 – 5:15 PM |
| Dr. Johannes Geith | Political Science | Big Tech and the State: Have Big Tech actors now accumulated state-like powers? How can we accurately measure authority of Big Tech corporations? And: What does the accumulation of authority, understood as “institutionalized forms or expressions of power,” mean for state-industry relations in the age of AI and in what ways do they differ across different locations? This SERC scholar group focuses on these—and related—questions, drawing on scholarship from a range of relevant research fields, in particular from political science, political economy, and economic history. A further key deliverable from this scholar group is to develop sophisticated skills in research methodology. During the first semester, the course reviews relevant literature systematically with the aim to facilitate students’ critical and analytical skills in reading and discussing academic research. Students are encouraged to select viable replication studies for developing their methodological toolkit. During the second semester, students develop research projects individually or in small groups. Deliverables may include short research papers, posters, or survey experiments. Students may forge partnerships with organizations such as the MIT GOV/LAB and are encouraged to develop creative measurement strategies or learn how to conduct survey experiments collaboratively. | Tuesdays, 2:00-3:00 PM |
| Dr. Ce Li | Laboratory for Information and Decision Systems (LIDS) | Game Theory, Economics, and AI: This reading group explores the growing interface of game theory, economics, and artificial intelligence. We study AI from two complementary perspectives: from agentic decision-making, with its capabilities such as information aggregation, recommendation, and forecasting; and from its reshaping of economics through the lens of incentives, learning, and economic outcomes. Topics of interests include, but not limited to, delegation to AI agents, benchmark design driven by mechanism design, preference elicitation and alignment, strategic behavior of AI agents, prediction markets under AI-driven forecasting, social learning in the era of AI, the economic and social implications of AI systems. Drawing on microeconomic theory, algorithmic game theory, and machine learning, the reading group aims to understand both the capabilities of increasingly autonomous AI systems and their broader implications for incentives, information, and economic outcomes. | Tuesdays, 2:45-4:00 PM |
| Dr. Ziv Epstein | Sloan | The Medium is the Mess: Today, algorithmic systems such as social media feeds and generative AI systems increasingly mediate human interactions and experiences. But the state of these systems is a Mess: the values embedded in these social algorithms lead to such outcomes as amplifying problematic content, inducing algorithmic overreliance and monoculture. Who determines what values are embedded in these systems? And what are their effects on human creativity and social fabrics? In this interdisciplinary group, we will excavate the deeper intentions and potentials underlying algorithmic objectives to design prosocial systems that support more creative and mindful interaction, for both producers and consumers of media. For producers, we will explore the impacts of AI on creativity and develop new theories, models and measurements for art in the age of algorithmic reproduction. For consumers, we will explore new prototypes of social media that give users more control and agency over their platform. | Wednesdays, 11:00 AM -12:30 PM |
| Dr. Erik Sandelin | Urban Studies and Planning | Anti/Micro/Techno/Fascism Design Studio: What if technology design was not fueled by a love of power? The term technofascism can be invoked to highlight the collusion of Big Tech and democratic backsliding or, more generally, to suggest various intersections between fascism and technology. Fascism is often understood as a historical and macropolitical phenomenon: the fascism of Hitler and Mussolini. But, as Foucault and Guattari remind us, there is also “the fascism in us all, in our heads and in our everyday behavior, the fascism that causes us to love power, to desire the very thing that dominates and exploits us”. This group explores, hands-on, the messy intersections of such microfascist desires and contemporary digital technologies. Through readings and design sessions, we practice gracious ways of creating technologies by not exercising all the force at our disposal. | Every third Wednesday, 1:00 PM – 4:00 PM |
| Dr. Bill Noble | Media Arts and Sciences | LLMs and Language as Commons: Large language Models (LLMs), the technology behind the current AI boom, are built on the social technology and collective cultural resource that is human language. Like other language technologies — the written word, the movable-type printing press, broadcast and social media, and others — LLMs will inevitably change language itself and our relationship to it. They already have. In this group, we’ll explore a variety of historical, linguistic, and technological topics that will give us tools to understand the changes taking place. We’ll discuss the Luddite movement of the early 1800s, which arose in response to market domination and labor abuses by English textile factor owners, and the enclosure of the commons that led up to that confrontation. Taking the technologically competent and socially aware perspective of the Luddite, we will learn about the material reality of the contemporary AI landscape: data centers, chips, consumer self-hosting and more. We will learn about how the tech sector’s construction of data as a commodity relates to AI and alternative models, such as the digital commons movement. Focusing on language itself, we’ll investigate the political causes and effects of language change, the global linguistic diversity crisis, linguistic alignment and meaning negotiation, and what all of this has to do with language technology. In the second half of the year, participants are invited to build on our investigations by devising a research project that probes the mechanics of language change and its relationship to language technologies, including conversational AI. I will offer guidance on methodologies from experimental, corpus-based, and computational linguistics, but students are encouraged to bring — and build on — their own interdisciplinary experiences. | Wednesdays, 1:30 – 2:45 PM |
| Dr. Ben Schwartz | Linguistics and Philosophy | Flourishing in a Digital Age: How can one live a good human life in our digital age? In this group, we will examine what it means to flourish in the context of recent and emerging digital technology. We will begin by considering some prominent accounts of human flourishing from the history of philosophy. Next, we will focus on the attention economy: a widespread market where people exchange their attention for access to a new media service (e.g., Instagram), who then sell this attention (and the data it generates) to advertisers. Here, we will explore ethical issues such as nudges, addiction, autonomy, corruption, and the value of attention. Then, we will examine related aspects of flourishing that are impacted by digital technology (including AI), such as relationships (e.g., dating apps, value capture, and relationships with AI), education, critical reflection, solitude, boredom, work, and creativity. Except for the section on relationships, these later topics will be chosen based upon the interests of the group members, given the time constraints of the weekly meetings. Some key questions that this group will examine include: Why is attention valuable? Is it morally permissible to treat attention as a commodity? What is valuable about relationships, education, thinking, solitude, work, creativity, etc., and what impact may recent and emerging digital technology have on these aspects of our lives? Students will be expected to study the primary readings before the weekly group meetings and attend each of these meetings ready to discuss this material. Students will also be expected to complete a final project. The format of this project is flexible: it could take the form of a final paper, but it could also be a blog post, podcast, etc. | Wednesdays, 2:00 – 3:15 PM |
| Dr. Erik Engberg | Sloan | AI, the Labor market, and Ethics: In this SERC study group, we will explore AI’s impact on the labor market and broader economy, from an ethical perspective. In our meetings we will discuss a couple of key topics within this area, with a bias toward the areas in which I have the most expertise, including: What impact has AI had on employment to date? What parts of the labor market are most affected? Should we be worried about large-scale job loss from AI? What are some key potential economic benefits of AI? When is it possible, or appropriate, to replace human interactions with AI? What is the best way to regulate new technology? How can increasingly autonomous AI agents be integrated into the economy and society? We will discuss relevant research, primarily from economics. Participants will get a sense of key questions, methods, data sources, and results in this literature. An overarching theme will be the question of how society can capture the potential benefits of AI, while minimizing the risks. We will discuss related ethical issues, highlighting some key tradeoffs. What is the right balance between preserving economic growth and development on the one hand, and protecting workers and consumers on the other? How do we regulate fast-moving technology in a way that mitigates risks, without needlessly stifling innovation? | Wednesdays, 3:00 – 4:00 PM |
| Dr. José Luis Gallegos-Quezada | Sloan | The Politics of AI: Who Wins, Who Pays, Who Decides?: Hollywood screenwriters went on strike over AI’s role in their work. Artists are suing companies that trained models on their creations. Data workers are demanding better pay, while communities are pushing back against data centers that strain local water and energy supplies. These debates reveal that AI is more than a technology: it is part of political and economic systems that create opportunities, costs, winners and losers. This scholar group will examine AI across its lifecycle: the resources it consumes, the data and labor behind it, its effects on workers and organizations, and the distribution of its economic gains. We will also consider its consequences for employment, inequality, democracy, and the environment. Three questions will guide us: Who benefits? Who pays? Who decides? Weekly discussions will connect big ideas from political philosophy and labor studies with current debates about responsible AI and regulation. Guest speakers from academia, government, industry, and social movements will introduce different perspectives and share firsthand experiences. Participants will turn their interests into two public-facing pieces in a format of their choice, such as a journalistic investigation, book review, interview, or op-ed. As a group, we will also help host a conference featuring leading voices in the field—an opportunity to meet experts, exchange ideas, and bring the semester’s conversations to a wider audience. | Wednesdays, 4:00 – 5:15 PM |
| Dr. Anna-Kaisa Kaila | MIT Media Lab | Resilient Creativity in the Age of AI: This SERC scholar group will explore how creative communities, artists, and cultural institutions are adapting to generative AI. How are the social, technical, and economic conditions of creative work impacted by synthetic content and AI-driven cultural production? What new kinds of artistic user communities, practices, and audiences are forming around contemporary AI tools? What concerns arise (e.g., around creative labor and agency, data ethics, or ownership), and what forms of counter-cultures, resistance, and community initiatives are emerging in response? During the Fall semester, we will meet weekly to discuss recent research and selected foundational texts on art and technology. Through close reading, practical exercises and taking turns to lead discussions, the participants will explore theoretical frameworks for analysis and develop their critical academic skills. In the Spring semester, participants begin to pursue an individual or group project that applies these perspectives to a research project, a case study, a policy proposal, or other creative deliverable. We approach the seminar topics mainly through the lenses of cultural studies, philosophy, and science and technology studies literature. Participants may choose to incorporate and evaluate contemporary generative AI tools as objects of critical analysis in their projects. While many of the examples discussed in the seminars will draw on music and sound, participants are encouraged focus in their projects on any creative domain that is aligned with their interests. | Thursdays, 10:00 – 11:00 AM |
| Dr. Jakob Stenseke | CSAIL | Aligning AI with the Better Version of You: Most work in AI alignment treats the task as making AI systems track what humans want –– their values, goals, interests, and preferences. But a large body of work holds that what people want under everyday conditions of incomplete information, time pressure, social influence, and cognitive load is frequently at odds with what they would want on reflection. Across AI and the broader attention economy, the former is systematically elicited and optimized for while the latter is squeezed out, with great costs to autonomy, well-being, and long-term goals. This project explores a flip in that framing. Instead of asking (only) how to make AI conform to what users want, we ask how users can be helped to reflect on and articulate their own conception of who they want to be—the better version of themselves—and how that articulation can be put back into the AI systems they use. Over the year we will develop, critique, and extend this idea from philosophical, psychological, technical, and governance angles, organized around three guiding questions: (1) What and whom is the better you? The conceptual and multidisciplinary foundations: what a more reflective version of a person consists in, who it answers to, and the philosophical and psychological support behind it. (2) Should we put the better you into the machine? Whether reflective values are the right alignment target in the first place, what the alignment and personalization literatures say, and the benefits and risks of doing this. (3) How do you put the better you into the machine? The technical and practical question: how to elicit reflective values and implement them across different “machines”, from AI assistants (via system prompts, fine-tuning, and steering) to recommender systems and attention feeds, together with the governance and monitoring this would require. | Thursdays, 2:00 – 3:00 PM |
| Dr. Alicja Ostrowska | History, Anthropology, and Science, Technology, and Society (HASTS) | How Does Programming Change Scientific Cultures?: New computational methods, such as AI, play an increasingly important role in science. However, they are not merely technological tools – they are practices that transform the ways in which we know the world. This reading group explores how computation changes the production of knowledge and what we understand as knowing. Drawing on classic works in Science and Technology Studies (STS), we will engage with questions of how science classifies the world by including and excluding certain objects. We will then delve into case studies from various domains – from biology to physics – to explore how programming changes scientific cultures. This course will provide a deeper understanding of how computation has transformed science historically and how it continues to reshape scientific practice in the age of AI. | Thursdays, 3:00 – 4:00 PM |
| Dr. Virgile Rennard | Political Science | Democracy and AI: Artificial intelligence is reshaping the core institutions of political life, how campaigns are run, how public opinion is measured, how governments respond to citizens, and how states compete for advantage. This reading group examines both how AI transforms politics, but also how Politics impact AI. Its aim is to pair the technical literacy needed to evaluate AI systems with the theoretical vocabulary of democratic thought. We ask how AI reconfigures participation, deliberation, and representation, and where it instead enables misinformation, surveillance, and authoritarian control. Topics include computational analysis of citizen input, the “liar’s dividend,” deliberative platforms and e-participation, bias in language models, algorithmic policing, and the competing regulatory frameworks emerging from the EU, the US, and China. | Thursdays, 3:00 – 4:15 PM |
| Dr. Yi-Hao Peng | CSAIL | Human-Centered Reinforcement Learning for Collaborative AI: My research builds collaborative, accessible AI agents that work alongside people, with a particular focus on those with disabilities. I develop data-driven models and interaction methods that help AI systems interpret multimodal context, collaborate with users proactively without undermining their agency, and define success in ways that reflect both shared standards and individual preferences. | TBD |