Center for Humans and Machines
Director: Iyad Rahwan
Introductory Overview
Guiding Concepts
Concept 1: Machine Behavior
Concept 2: Science Fiction Science
Concept 3: Machine Culture
Introductory Overview
The Center for Humans and Machines (CHM) conducts interdisciplinary science to understand, anticipate, and shape major disruptions from digital media and artificial intelligence (AI) to the way we think, learn, work, play, cooperate, and govern.
The Center's vision is to be a world-leading hub of scientific knowledge about the current and future impact of AI and digitalization on human society, economy, and culture, thus informing both system design and policy-making. CHM believes the challenges posed by the information revolution are no longer mere computer science problems, but require deep integration of computational techniques with methods from across the behavioral sciences.
In the following, the Center reports on work carried out between 2023 and 2026. This report describes various projects, organized into broad themes, which are diverse in terms of their scientific methodology and the research questions they explore. Each theme has an overarching research question, as shown in Figure 1. First, the report outlines the overall scientific approach of CHM and its relationship to existing disciplines. This will be followed by a selection of completed and ongoing research projects in each theme that will be described in detail below.
Guiding Concepts
The Center is distinguished by the following three guiding concepts that help identify and scope research questions. The Center has also been instrumental in the conceptual development and ongoing popularization of some of these concepts within the scientific community.
Understanding machine behavior, including human perception and reaction to such behavior, requires concepts and methodologies from across the behavioral sciences.
The main focus of the Center is around the notion of machine behavior, with two broad aspects of interest: (1) how intelligent machines behave, and the outcomes that emerge as machines interact with humans; and (2) how humans perceive the behavior of machines, and how this perception shapes their expectations and judgment of the machines' actions and their own behavior. The contours of the emerging field of machine behavior were outlined in "Machine behavior" (Rahwan et al., 2019).
Despite fundamental differences between machines and biological organisms, CHM draws inspiration from Nikolaas Tinbergen's four questions of biology in order to organize the different kinds of questions one might ask about machine behavior. Machines have mechanisms that produce behavior, undergo development that integrates environmental information into behavior, produce functional consequences that cause specific machines to become more or less common in specific environments, and embody evolutionary histories through which past environments and human decisions continue to influence machine behavior. These four levels of analysis are summarized in Figure 2.
Fundamental questions in machine behavior include the emergence of human–machine cooperation, potential social dilemmas and moral hazards that may arise from human–machine interaction, the potential role of machines as social catalysts, and the impact of AI on human culture.
Adapted from Rahwan et al. (2019)
Key References
Concept 2: Science Fiction Science
To anticipate the impact of future technologies on humans, the Center combines imagination of possible futures with a scientific approach to studying behavior.
Regulators and the momentum of emerging technologies face the so-called Collingridge dilemma (Collingridge, 1982):
- Information Paradox: Early in development, technology is easy to change but its potential impact is unknown, making control difficult.
- Control Paradox: Once mature, consequences are clear, enabling informed control, but the technology is entrenched and hard to regulate.
One way to mitigate the dilemma is to make regulation more responsive and technology design more agile. But, in addition, can we improve our ability to predict the societal impact of technologies?
Technological forecasting is already a well-established field. But it is largely based on qualitative methods (e.g., Delphi method to elicit expert opinion about technology readiness) or computer simulation (e.g., of the dynamics of adoption of autonomous vehicles). But these approaches lack the behavioral measurement and experimental control characteristics of behavioral science.
Science Fiction Science (Sci-Fi-Sci) (Rahwan et al., 2025), a concept developed at the Center, immerses humans in different simulated futures—each representing a distinct version of a future technology design or regulatory framework. People's reactions, beliefs, and behaviors are then observed, analyzed, and tested. This allows us to anticipate problems, informing policymaking and design, before the technology is fully deployed.
Rahwan et al. (2025)
This approach has already yielded impactful foresight, from the power of social media to mobilize people at scale (Pickard et al., 2011), the ethical dilemmas facing the regulation and design of self-driving cars (Awad et al., 2019), to the likely increase in tax fraud when humans can delegate tasks to AI agents (Köbis, Rahwan, et al., 2025). In an era of accelerating technological progress, the Sci-Fi-Sci method holds enormous potential to help us prepare, rather than simply react, to technology's impact on humans.
Science fiction author Isaac Asimov said, "A good science-fiction story should be able to predict not the automobile but the traffic jam." The Center's goal is to predict the traffic jams of future digital technologies.
Key References
Collingridge, D. (1982) The social control of technology. St. Martin's Press.
The future will be determined by evolutionary dynamics among superminds: groups of humans and AI systems jointly building and refining cultural artefacts and institutions for innovation, cooperation, and coordination.
The term culture describes information that affects the behavior of individuals, who acquire it from other members of their species through teaching, imitation, and other forms of social transmission. It is now increasingly recognized that humans have become the dominant species on Earth not through their individual intelligence, but through their ability to create and transmit adaptive cultural innovations (norms, institutions, technological know-how, etc.) through a process of cultural evolution (Henrich, 2016).
The Information Age is facilitating two monumental changes in human organization and culture. First, digital communication technologies, such as smartphones and social media, are massively expanding our ability to organize and coordinate at scale. Second, creative software is radically changing how we create, store, and share cultural information (Brinkmann et al., 2023). AI tools are facilitating new kinds of hybrid human–machine creative practices. Recommender systems are shaping how cultural information is disseminated. Generative AI software is even becoming an active "agent" in the culture generation process, via computer-generated artistic and technological inventions.
These two phenomena—cultural evolution and the information revolution—give primacy to emerging forms of human–machine collective intelligence. Consequently, it is likely that future human progress will be shaped by competition operating not among individual humans or groups of humans, but between hybrid human–machine collectives, or superminds (Malone, 2018).
Key References
Henrich, J. (2016). The secret of our success: How culture is driving human evolution, domesticating our species, and making us smarter. Princeton University Press.
Malone, T. W. (2018). Superminds: The surprising power of people and computers thinking together. Little, Brown Spark.