Forum on Emerging Technology

From Interface to Agency
A New Discourse for Design and AI
Grimes Engineering Center
University of California, Berkeley
Friday, March 20, 2026
9:00 AM – 5:30 PM, with reception to follow
The 2026 MDes Forum on Emerging Technologies examines an emerging artifact of design – the Agentic System – and interrogates the theories, practices, methods, and histories that are required to attend to it effectively, ethically, and critically. We propose to call this field Agent Design or Design for Agentic Systems.
More about Agent Design, and the fields we’re convening…
We understand “Agentic Systems” broadly as a new class of intelligent, persistent, and adaptive system, and as any network of computational actors capable of autonomous perception and action within their environment. We assert that designers attending to such ecologies must shift their mindset away from interfaces and environments, and toward domains of action. Further, they must critically examine their responsibility to those whom these Agents claim to act on behalf of. The still-nascent Agents being designed today – the chatbots, assistants, and recommenders – are described, often naively, as systems that act on behalf of users. But in practice, these Agents are more accurately understood as acting on behalf of the corporations or governments that built them, often with little transparency or accountability. As such, the phrase “on behalf of” must be met critically, and with open eyes. It must be interrogated. It must be expanded to include a broader range of constituencies, stakeholders, and publics. Since we see that Agentic Systems must be treated as consequential actors and not neutral tools, we seek to ask: what is automated, for whom, and at what cost? By naming alternative frontiers for agentic practice that better serve the public good, we resist the seemingly-inevitable tendency of today’s Agents to further privatize and enclose the commons.
Since we see that Agentic Systems must be treated as consequential actors and not neutral tools, we seek to ask: what is automated, for whom, and at what cost? By naming alternative frontiers for agentic practice that better serve the public good, we resist the seemingly-inevitable tendency of today’s Agents to further privatize and enclose the commons. To this end, we seek to shift our understanding of Agents from models of inter-action to frameworks of action-on-behalf-of.
The Berkeley MDes Forum seeks to convene adjacent fields for what each contributes to this larger project: cybernetics (feedback and adaptation, learning, conversation); ontological design (how language / tools configure worlds and roles); critical making (argument through artifact); critical data studies (provenance, power, privacy); generative design (model-driven authorship); service design (operational venues across stakeholders); and science and technology studies (sociotechnical context and policy).
The design of Agentic Systems is already occurring “in the wild”. But this work is not yet a field. At some point, a particular constellation of methods, theories, and histories will crystallize and sustain a formal practice that attends to the design of Intelligent Agents. From that future vantage point, this constellation will seem obvious and its configuration will appear inevitable. Absent this clarity, today we are left only to speculate on the nature of design practice that does not yet fully exist. And yet, to guess is to give shape. The Forum identifies a threshold condition in technological and cultural practice – an inflection point not yet stabilized into a field – and brings together designers, scholars, and practitioners to examine it before it hardens into convention.
Less
Keynote Speakers
Andrew Pickering
Andrew Pickering is emeritus professor in the Department of Social and Political Sciences, Philosophy, and Anthropology at Exeter University. He is internationally known as a leader in the field of science and technology studies. He is the author of several books including The Mangle of Practice: Time, Agency and Science and The Cybernetic Brain: Sketches of Another Future. His latest book, Acting with the World: Agency in the Anthropocene (Duke UP, 2025), explores neo-cybernetic engagements with the environment, including water engineering, rewilding, farming, wildfires and animism. He lives in the UK.
more about Pickering, and what he might contribute…
Pickering brings a foundational theoretical frame for understanding agency through cybernetics, emphasizing systems that adapt, learn, and act in unfolding relations with their environments. His perspective situates today’s agents within longer histories of performative, non-deterministic systems, offering conceptual grounding for designers navigating emergent ecologies of action and responsibility.
Less
Maggie Gram
Maggie Gram is a writer, cultural historian, and designer. She is the author of The Invention of Design: A Twentieth Century History (2025), and has also written for n+1 and the New York Times. She leads an experience-design team at Google, and has taught at the Maryland Institute College of Art, Washington University in St. Louis, Ca’ Foscari University of Venice, and Harvard University. She lives in New York.
more about Gram, and what she might contribute…
Gram offers a historian’s and designer’s view of how shifts in tools, rhetoric, and institutional arrangements repeatedly remake what design is and does. Her contribution helps situate agentic systems within broader transformations of design practice, illuminating how disciplines evolve and how pedagogy, ethics, and professional norms must adapt to new technological actors.
Less
Liz Danzico
Liz Danzico is a designer and educator at the intersection of strategy, product, and innovation. As VP of Design for Microsoft AI, she leads design for products that reach billions – from Bing Search to Copilot – advancing responsible AI to empower people. As Founding Chair of the MFA Interaction Design program at SVA, she has shaped the careers of graduates now leading design at top global companies. She served as SVP of Digital at NPR, helping to launch award-winning NPR One, NPR apps, and strategic planning processes, using design as a business tool. Liz advises organizations that amplify underrepresented voices. She lives in New York.
more about Danzico, and what she might contribute…
Danzico brings a deeply human-centered and practice-oriented sensibility to the design of emerging systems, emphasizing collaboration, communication, and care over control. Her perspective highlights how designers shape relationships between people and technologies, and how attentiveness, trust, and shared agency become essential materials when autonomous systems act within complex organizational and cultural contexts.
Less
Bruce Sterling
Bruce Sterling is an American science fiction author known for his novels and short fiction. In particular, he is linked to the cyberpunk subgenre. He is also a design critic and has contributed to design discourse, including work on “design fiction” and his book, Shaping Things (2005). He is Professor of New Media and Science Fiction at The European Graduate School / EGS. He lives in Ibiza.
more about Sterling, and what he might contribute…
Sterling contributes a cultural and speculative lens, drawing from fiction, design criticism, and media theory to reveal how imagined futures shape real technologies. His perspective underscores the narrative and conceptual inheritances embedded in today’s agentic systems, helping participants see how past visions, myths, and prototypes quietly inform what – and whom – our systems now act for.
Less
Dialogs
Agent Design as Ontological Design
Terry Winograd
Terry Winograd is Professor Emeritus at Stanford University, and founder of the Stanford HCI Group; He is the author of ‘Understanding Computers and Cognition‘.
more about Winograd, and what he might contribute…
Winograd provides historical depth from early work in AI, language, and interaction, showing how formative ideas about conversation, commitments, and situated action helped establish an ontological design tradition – one in which technologies shape the meanings, roles, and practices available to their users. His perspective clarifies how these foundational commitments continue to structure today’s agentic systems, and which assumptions must be reexamined as agents become persistent actors in lived environments.
Less
Andy Matuschak
Andy Matuschak is an independent researcher. He helped build iOS at Apple and led R&D at Khan Academy.
more about Matuschak, and what he might contribute…
Matuschak brings a contemporary view of agents as partners in cognition, learning, and long-term knowledge work, demonstrating how modern systems actively participate in shaping attention, understanding, and everyday practice. His contribution highlights the ongoing relevance of ontological design: how agents remake the conditions under which thinking and doing unfold, and how their behaviors must be carefully crafted to support meaningful, reflective relationships between humans and computational actors.
Less
Taylor Rogalski
Taylor Rogalski is a designer working at the intersection of education, games, and computing as a medium. He is working on Pico – an agentic approach to understanding screenshots – and also designed the RLHF system used to train Claude.
more about Rogalski, and what he might contribute…
Rogalski brings a builder’s perspective on the infrastructures that make contemporary agents behave, and the concrete design choices about what agents notice, remember, prioritize, and ultimately do on behalf of their users.
Less
Crafting Interactions for Agentic Systems
Jodi Forlizzi
Jodi Forlizzi is the Herbert A. Simon Professor of Computer Science and HCII at Carnegie Mellon University, where she studies the ethical impacts of human interaction with AI systems.
more about Forlizzi, and what she might contribute…
Forlizzi offers a design-research lens on behavior, embodiment, and interaction, emphasizing how systems are shaped by – and in turn shape – human routines and expectations. Her contribution clarifies how designers can operationalize concepts like legibility, predictability, and social cues to craft agentic systems that behave responsibly and meaningfully within everyday environments.
Less
Wendy Ju
Wendy Ju is Associate Professor at Cornell Tech, and is the author of ‘The Design of Implicit Systems‘. She studies how interactive devices can engage people without intruding upon them.
more about Ju, and what she might contribute…
Ju brings expertise in human–robot interaction and situated behavior, highlighting how subtle actions, timing, and spatial cues influence perception and trust in autonomous systems. Her insights help participants understand how agents communicate through motion and presence, and how interaction design principles translate into reliable, accountable behavior in complex real-world settings.
Less
Ivan Poupyrev
Ivan Poupyrev is the Founder and CEO of Archetype AI – a “physical AI company” that seeks to understand the physical world from sensor data. He is a National Design Award Winner and formerly held roles at Google and Disney Imagineering.
more about Poupyrev, and what he might contribute…
Poupyrev provides an applied, industry-facing view of agentic systems built from continuous sensing, physical-world modeling, and real-time adaptive behavior. His work illustrates how interfaces dissolve into ambient intelligence, raising concrete questions of power, authorship, accountability, and service operations. He grounds the conversation in what is technically feasible today and what infrastructural agents may soon become.
Less
Critical Spatial Practices for Agentic Systems
Eryk Salvaggio
Eryk Salvaggio is a Gates Scholar at the University of Cambridge for a PhD in the Digital Humanities. As a researcher and media artist, he is interested in the social and cultural impacts of artificial intelligence.
more about Salvaggio, and what he might contribute…
Salvaggio contributes a critical data studies perspective grounded in artistic and research practice, revealing how archives, images, and built environments become datasets that structure agentic perception and action. His work exposes the ideological gaps between datasets and lived spaces, offering methods for contesting and reshaping the computational imaginaries embedded within spatial systems.
Less
Catherine Griffiths
Catherine Griffiths is an Assistant Professor at the Columbia Graduate School of Architecture, Planning and Preservation (GSAPP). She researches on the spatial politics of AI.
more about Griffiths, and what she might contribute…
Griffiths explores how algorithmic systems reorganize labor, governance, and rights across homes, workplaces, and cities, treating AI as a political and spatial force. Her research and creative practice make visible the hidden infrastructures and power asymmetries encoded in machine learning, offering speculative and documentary strategies for challenging normative design logics and imagining counter-algorithmic futures.
Less
Andrew Witt
Andrew Witt is the inaugural Kavita and Krishna Bharat Professor of AI and Design at Washington University, and the founder of Certain Measures.
more about Witt, and what he might contribute…
Witt brings a systems-oriented architectural and computational design perspective grounded in the work of Certain Measures, where physical and digital worlds are fused through data-rich spatial design. His contribution highlights how agents perceive, interpret, and reshape built environments, demonstrating how computational modeling, simulation, and software enable new forms of spatial intelligence and infrastructural agency across cultural, industrial, and civic contexts.
Less
Lightning Talks
Short cases or demonstrations that foreground deployed or deployable agents in public or institutional contexts.
Includes:
- TJ McLeish, Lecturer, UC Berkeley
- Gray Crawford, Product Designer
- Max Kreminski, Faculty, Cornell
- John Cain, founder of Zerowidth
- Steve Brumby, founder and CEO of Impact Observatory
- Weiwei Xu, founder of AI start-up Sprout Place