@article{RauwolfBryson, author = {Rauwolf, Paul and Bryson, Joanna}, title = {Expectations of Fairness and Trust Co-Evolve in Environments of Partial Information}, series = {Dynamic Games and Applications}, volume = {8}, journal = {Dynamic Games and Applications}, number = {4}, doi = {10.1007/s13235-017-0230-x}, pages = {891 -- 917}, abstract = {When playing one-shot economic games, individuals often blindly trust others, accepting partnerships without any information regarding the trustworthiness of their partner. Consequently, they risk deleterious pacts. Oddly, when individuals do have information about another, they reject partnerships that are not fair, despite the fact that such offers are profitable—individuals costly punish. Why would one reject profitable partnerships on the one hand, but risk unknown offers on the other? Significant research has gone into explaining the contexts where blind trust or costly punishment provides an evolutionary advantage; however, both behaviours are rarely considered in tandem. Here we demonstrate that both behaviours can simultaneously be revenue maximizing. Further, given the plausible condition of partially obscured information and partner choice, trust mediates the generation of costly punishment. This result is important because it demonstrates that the evolutionary viability of trust, fairness, and costly punishment may be linked. The adaptive nature of fairness expectations can best be explained in concert with trust.}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {Patiency Is Not a Virtue: The Design of Intelligent Systems and Systems of Ethics}, series = {Ethics and Information Technology}, volume = {20}, journal = {Ethics and Information Technology}, number = {1}, doi = {10.1007/s10676-018-9448-6}, pages = {15 -- 26}, abstract = {The question of whether AI systems such as robots can or should be afforded moral agency or patiency is not one amenable either to discovery or simple reasoning, because we as societies constantly reconstruct our artefacts, including our ethical systems. Consequently, the place of AI systems in society is a matter of normative, not descriptive ethics. Here I start from a functionalist assumption, that ethics is the set of behaviour that maintains a society. This assumption allows me to exploit the theoretical biology of sociality and autonomy to explain our moral intuitions. From this grounding I extend to consider possible ethics for maintaining either human- or of artefact-centred societies. I conclude that while constructing AI systems as either moral agents or patients is possible, neither is desirable. In particular, I argue that we are unlikely to construct a coherent ethics in which it it is ethical to afford AI moral subjectivity. We are therefore obliged not to build AI we are obliged to.}, language = {en} } @techreport{WilsonBrysonTheodorou, type = {Working Paper}, author = {Wilson, Holly and Bryson, Joanna and Theodorou, Andreas}, title = {Perceptions of Moral Dilemmas in a Virtual Reality Car Simulation}, edition = {Paper presented at IA Symposium}, abstract = {The prevalence of artificial intelligent agents carrying out morally salient decisions is growing. The decisions made by such agents as autonomous cars or weapon systems, may have life and death consequences. We argue that the decision-making algorithms of all agents whose decisions have high societal impact should be transparent [6]; to ensure human-agent interaction is fully informed, consensual, and of maximum benefit to the society. Importantly, the literature also indicates we may perceive and respond to morally salient decisions made by a machine differently to the same decision made by a human [5, 4, 3]. We present here a virtual reality simulation of a self-driving car we developed, in which users experience moral dilemmas. In our two studies, we investigate the perceptions of a morally salient decision; first as moderated by the type of the agent, artificial or natural (human), and then with the implementation of transparency. Specifically, inspired by the Moral Machine research programme [2, 1], we used social value as a moral framework. The agent chooses to hit a pedestrian on either the left or right side of a zebra crossing dependent on dimensions of occupation, body-size and gender. Participants gave feedback after each scenario. Contrary to past findings, participants in this current study were distressed at the principle of decision-making based on attributes such as social value. In questionnaire responses and postexperiment conversation the majority reported preferring such decisions to be made at random. This raises important insights into how we implement moral frameworks. We suggest that the disparity between preferences from the current study and past work is due to the virtual reality methodology we used. Specifically, we note a distinction between emotional vs. rational decisionmaking, which was supported by an extension survey we conducted. Consistent with expectations, the self-driving car was perceived as significantly less morally culpable and human-like than the human-driver. The transparency implementation led to a further significant reduction in perceived human-likeness, and also to reduced perceptions of intentionality. The reduction in moral culpability has disturbing possible connotations, though it may also be helpful for correct attribution of accountability. Promisingly, our transparency implementation significantly improved participants' understanding of the self-driving cars decision. We suggest companies implementing moral frameworks do not take crowd-sourced preferences at face value, but explore the methodology used. Additionally, our work supports transparency as a mechanism to calibrate our mental models of autonomous agents.}, language = {de} } @article{GaudlBryson, author = {Gaudl, Swen E. and Bryson, Joanna}, title = {The extended ramp model: A biomimetic model of behaviour arbitration for lightweight cognitive architectures}, series = {Cognitive Systems Research}, volume = {50}, journal = {Cognitive Systems Research}, doi = {10.1016/j.cogsys.2018.02.001}, pages = {1 -- 9}, abstract = {In this article, we present an idea for a more intuitive, low-cost, adjustable mechanism for behaviour control and management. One focus of current development in virtual agents, robotics and digital games is on increasingly complex and realistic systems that more accurately simulate intelligence found in nature. This development introduces a multitude of control parameters creating high computational costs. The resulting complexity limits the applicability of AI systems. One solution to this problem it to focus on smaller, more manageable, and flexible systems which can be simultaneously created, instantiated, and controlled. Here we introduce a biologically inspired systems-engineering approach for enriching behaviour arbitration with a low computational overhead. We focus on an easy way to control the maintenance, inhibition and alternation of high-level behaviours (goals) in cases where static priorities are undesirable. The models we consider here are biomimetic, based on neuro-cognitive research findings from dopaminic cells responsible for controlling goal switching and maintenance in the mammalian brain. The most promising model we find is applicable to selection problems with multiple conflicting goals. It utilizes a ramp function to control the execution and inhibition of behaviours more accurately than previous mechanisms, allowing an additional layer of control on existing behaviour prioritization systems.}, language = {en} } @incollection{WilsonRauwolfBryson, author = {Wilson, Holly and Rauwolf, Paul and Bryson, Joanna}, title = {Evolutionary Psychology and Artificial Intelligence: The Impact of Artificial Intelligence on Human Behaviour}, series = {The SAGE Handbook of Evolutionary Psychology}, booktitle = {The SAGE Handbook of Evolutionary Psychology}, editor = {Shackelford, Todd K.}, publisher = {SAGE Publications Ltd}, address = {London}, isbn = {9781526489166}, pages = {27}, abstract = {Artificial Intelligence (AI) presents a new landscape for humanity. Both what we can do, and the impact of our ordinary actions is changed by the innovation of digital and intelligent technology. In this chapter we postulate how AI impacts contemporary societies on an individual and collective level. We begin by teasing apart the current actual impact of AI on society from the impact that our cultural narratives surrounding AI has. We then consider the evolutionary mechanisms that maintain a stable society such as heterogeneity, flexibility and cooperation. Taking AI as a prosthetic intelligence, we discuss how—for better and worse—it enhances our connectivity, coordination, equality, distribution of control and our ability to make predictions. We further give examples of how transparency of thoughts and behaviours influence call-out culture and behavioural manipulation with consideration of group dynamics and tribalism. We next consider the efficacy and vulnerability of human trust, including the contexts in which blind trust in information is either adaptive or maladaptive in an age where the cost of information is decreasing. We then discuss trust in AI, and how we can calibrate trust as to avoid over-trust and mistrust adaptively, using transparency as a mechanism. We then explore the barriers for AI increasing accuracy in our perception by focusing on fake news. Finally, we look at the impact of information accuracy, and the battles of individuals against false beliefs. Where available, we use models drawn from scientific simulations to justify and clarify our predictions and analysis.}, language = {en} } @incollection{BrysonTheodorou, author = {Bryson, Joanna and Theodorou, Andreas}, title = {How Society Can Maintain Human-Centric Artificial Intelligence}, series = {Human-Centered Digitalization and Services}, booktitle = {Human-Centered Digitalization and Services}, editor = {Toivonen, Marja and Saari, Eveliina}, publisher = {Springer}, isbn = {978-981-13-7725-9}, publisher = {Hertie School}, pages = {305 -- 323}, abstract = {Although not a goal universally held, maintaining human-centric artificial intelligence is necessary for society's long-term stability. Fortunately, the legal and technological problems of maintaining control are actually fairly well understood and amenable to engineering. The real problem is establishing the social and political will for assigning and maintaining accountability for artifacts when these artefacts are generated or used. In this chapter we review the necessity and tractability of maintaining human control, and the mechanisms by which such control can be achieved. What makes the problem both most interesting and most threatening is that achieving consensus around any human-centred approach requires at least some measure of agreement on broad existential concerns.}, language = {en} } @inproceedings{RotsidisTheodorouBrysonetal., author = {Rotsidis, Alexandros and Theodorou, Andreas and Bryson, Joanna and Wortham, Robert H.}, title = {Improving Robot Transparency: An Investigation With Mobile Augmented Reality}, series = {Paper presented at The 28th IEEE International Conference on Robot \& Human Interactive Communication, New Delhi, India}, booktitle = {Paper presented at The 28th IEEE International Conference on Robot \& Human Interactive Communication, New Delhi, India}, pages = {8}, abstract = {Autonomous robots can be difficult to understand by their developers, let alone by end users. Yet, as they become increasingly integral parts of our societies, the need for afford- able easy to use tools to provide transparency grows. The rise of the smartphone and the improvements in mobile computing performance have gradually allowed Augmented Reality (AR) to become more mobile and affordable. In this paper we review relevant robot systems architecture and propose a new software tool to provide robot transparency through the use of AR technology. Our new tool, ABOD3-AR provides real-time graphical visualisation and debugging of a robot's goals and priorities as a means for both designers and end users to gain a better mental model of the internal state and decision making processes taking place within a robot. We also report on our on-going research programme and planned studies to further understand the effects of transparency to naive users and experts.}, language = {en} } @article{WorthamGaudlBryson, author = {Wortham, Robert H. and Gaudl, Swen E. and Bryson, Joanna}, title = {Instinct: A Biologically Inspired Reactive Planner for Intelligent Embedded Systems}, series = {Cognitive Systems Research}, volume = {57}, journal = {Cognitive Systems Research}, doi = {10.1016/j.cogsys.2018.10.016}, pages = {207 -- 215}, abstract = {The Instinct Planner is a new biologically inspired reactive planner, based on an established behaviour based robotics methodology and its reactive planner component—the POSH planner implementation. It includes several significant enhancements that facilitate plan design and runtime debugging. It has been specifically designed for low power processors and has a tiny memory footprint. Written in C++, it runs eciently on both Arduino(Atmel AVR) and Microsoft VC++ environments and has been deployed within a low cost maker robot to study AI Transparency. Plans may be authored using a variety of tools including a new visual design language, currently implemented using the Dia drawing package.}, language = {en} } @incollection{Bryson, author = {Bryson, Joanna}, title = {The Artificial Intelligence of the Ethics of Artificial Intelligence: An Introductory Overview for Law and Regulation}, series = {The Oxford Handbook of Ethics of AI}, booktitle = {The Oxford Handbook of Ethics of AI}, editor = {Dubber, Markus and Pasquale, Frank and Das, Sunit}, publisher = {Oxford University Press}, isbn = {9780190067397}, doi = {10.1093/oxfordhb/9780190067397.013.1}, publisher = {Hertie School}, pages = {1000}, abstract = {Artificial intelligence (AI) is a technical term often referring to artifacts used to detect contexts for human actions, or sometimes also for machines able to effect actions in response to detected contexts. Our capacity to build such artifacts has been increasing, and with it the impact they have on our society. This does not alter the fundamental roots or motivations of law, regulation, or diplomacy, which rest on persuading humans to behave in a way that provides sustainable security for humans. It does however alter nearly every other aspect of human social behaviour, including making accountability and responsibility potentially easier to trace. This chapter reviews the nature and implications of AI with particular attention to how they impinge on possible applications to and of law.}, language = {en} } @incollection{Bryson, author = {Bryson, Joanna}, title = {The Past Decade and Future of AI's Impact on Society}, series = {Towards a New Enlightenment? A Transcendent Decade}, volume = {11}, booktitle = {Towards a New Enlightenment? A Transcendent Decade}, publisher = {BBVA}, isbn = {9788417141219}, publisher = {Hertie School}, abstract = {Artificial intelligence (AI) is a technical term referring to artifacts used to detect contexts or to effect actions in response to detected contexts. Our capacity to build such artifacts has been increasing, and with it the impact they have on our society. This article first documents the social and economic changes brought about by our use of AI, particularly but not exclusively focusing on the decade since the 2007 advent of smartphones, which contribute substantially to "big data" and therefore the efficacy of machine learning. It then projects from this political, economic, and personal challenges confronting humanity in the near future, including policy recommendations. Overall, AI is not as unusual a technology as expected, but this very lack of expected form may have exposed us to a significantly increased urgency concerning familiar challenges. In particular, the identity and autonomy of both individuals and nations is challenged by the increased accessibility of knowledge.}, language = {en} } @book{TheodorouBrysonBandtLaw, author = {Theodorou, Andreas and Bryson, Joanna and Bandt-Law, Bryn}, title = {The Sustainability Game: AI Technology as an Intervention for Public Understanding of Cooperative Investment}, series = {IEEE CONFERENCE ON GAMES (COG)[8848058] IEEE}, journal = {IEEE CONFERENCE ON GAMES (COG)[8848058] IEEE}, isbn = {9781728118840}, publisher = {Hertie School}, abstract = {Cooperative behaviour is a fundamental strategy for survival; it positively affects economies, social relationships, and makes larger societal structures possible. People vary, however, in their willingness to engage in cooperative behaviour in a particular context. Here we examine whether AI can be effectively used to to alter individuals' implicit understanding of cooperative dynamics, and hence increase cooperation and participation in public goods projects. We developed an intervention---the Sustainability Game (SG)---to allow players to experience the consequences of individual investment strategies on a sustainable society. \%, when personal well being, communal space, and resources limitations are taken into consideration. Results show that the intervention significantly increases individuals' cooperative behaviour in partially anonymised public goods contexts, but enhances competition one-on-one. This indicates our intervention does improve transparency of the systemic consequences of individual cooperative behaviour.}, language = {en} } @incollection{WorthamBryson, author = {Wortham, Robert H. and Bryson, Joanna}, title = {Communication}, series = {Living Machines: A Handbook of Research in Biomimetic and Biohybrid Systems}, booktitle = {Living Machines: A Handbook of Research in Biomimetic and Biohybrid Systems}, editor = {Prescott (et al.), Tony J.}, isbn = {9780199674923}, doi = {10.1093/oso/9780199674923.003.0033}, publisher = {Hertie School}, pages = {312 -- 326}, abstract = {From a traditional engineering perspective, communication is about effecting control over a distance, and its primary concern is the reliability of transmission. This chapter reviews communication in nature, describing its evolution from the perspective of the selfish gene. Communication in nature is ubiquitous and generally honest, and arises as much from collaboration as manipulation. We show that context and relevance allow effective communication with little information transfer, particularly between organisms with similar capacities and goals. Human language differs fundamentally from the non-verbal communication we share with other animals; robots may need to accommodate both. We document progress in AI capacities to generate synthetic emotion and to sense and classify human emotion. Communication in contemporary biomimetic systems is between robots in swarm robotics, but also between robot and human for both autonomous and collaborative systems. We suggest increased future emphasis on capacities to receive and comprehend signs, and on the pragmatic utility of communication and cooperation.}, language = {en} } @techreport{BrysonStewartMcCarty, type = {Working Paper}, author = {Bryson, Joanna and Stewart, Alexander J. and McCarty, Nolan}, title = {Polarization under rising inequality and economic decline}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:b1570-opus4-36522}, abstract = {Social and political polarization is a significant source of conflict and poor governance in many societies. Understanding its causes has become a priority of scholars across many disciplines. Here we demonstrate that shifts in socialization strategies analogous to political polarization and identity politics can arise as a locally-beneficial response to both rising wealth inequality and economic decline. Adopting the perspective of cultural evolution, we develop a framework to study the emergence of polarization under shifting economic environments. In many contexts, interaction with diverse out-groups confers benefits from innovation and exploration greater than those that arise from interacting exclusively with a homogeneous in-group. However, when the economic environment favors risk-aversion, a strategy of seeking lower-risk interactions can be important to maintaining individual solvency. To capture this dynamic, we assume that in-group interactions have a lower expected outcome than out-group interactions, but also one that is more certain. Thus in-group interactions are less risky than out-group interactions. Our model shows that under conditions of economic decline or increasing wealth inequality, some members of the population benefit from adopting a risk-averse, in-group favoring strategy. Moreover, we show that such in-group polarization can spread rapidly to the whole population and persist even when the conditions that produced it have reversed. Finally we offer empirical support for the role of income inequality as a contributor to affective polarization in the United States. Our work provides a framework for studying how disparate forces interplay, via cultural evolution, to shape patterns of identity, and unifies what are often seen as conflicting explanations for political polarization: identity threat and economic anxiety.}, language = {en} } @article{StewartMcCartyBryson, author = {Stewart, Alexander J. and McCarty, Nolan and Bryson, Joanna}, title = {Polarization under rising inequality and economic decline}, series = {Science Advances}, volume = {6}, journal = {Science Advances}, number = {50}, doi = {10.1126/sciadv.abd4201}, abstract = {Social and political polarization is an important source of conflict in many societies. Understanding its causes has become a priority of scholars across disciplines. We demonstrate that shifts in socialization strategies analogous to political polarization can arise as a locally beneficial response to both rising wealth inequality and economic decline. In many contexts, interaction with diverse out-groups confers benefits from innovation and exploration greater than those that arise from interacting exclusively with a homogeneous in-group. However, when the economic environment favors risk aversion, a strategy of seeking lower-risk in-group interactions can be important to maintaining individual solvency. Our model shows that under conditions of economic decline or increasing inequality, some members of the population benefit from adopting a risk-averse, in-group favoring strategy. Moreover, we show that such in-group polarization can spread rapidly to the whole population and persist even when the conditions that produced it have reversed.}, language = {en} } @article{BrysonMalikova, author = {Bryson, Joanna and Malikova, Helena}, title = {Is There an AI Cold War?}, series = {Global Perspectives}, volume = {2}, journal = {Global Perspectives}, number = {1}, doi = {10.1525/gp.2021.24803}, abstract = {Regulation is a means societies use to create the stability, public goods, and infrastructure they need to thrive securely. This policy brief is intended to both document and to address claims of a new AI cold war: a binary competition between the United States and China that is too important for other powers to either ignore or truly participate in directly, beyond taking sides. We argue that while some of the claims of this narrative are based at least in part on genuine security concerns and important unknowns, evidence for its extreme binary nature is lacking. This absence of factual evidence is concerning, because related geopolitical tensions may be used to interfere with regulation of AI and agencies associated with its development. Here we first document and then analyze the extremely bipolar picture prominent policymakers and political commentators have been recently painting of the AI technological situation, portraying China and the United States as the only two global powers. We then examine the plausibility of these claims using two measures: internationally registered AI patents and the market capitalization of the companies that hold them. These two measures, while each somewhat arbitrary and imperfect, are often deployed in the context of the binary narrative and can therefore be seen as conservative choices in that they should favor exactly the "champions" of that narrative. In fact, these measures do not produce bipolar results: Chinese capacity has been exaggerated and that of other global regions deprecated. These findings call into question the motivation behind the documented claims, though they also further illuminate the uncertainty concerning digital technology security. We recommend that all parties engage in contributing to a safe, secure, and transparent regulatory landscape.}, language = {en} } @incollection{BrysonEisenlauer, author = {Bryson, Joanna and Eisenlauer, Martin}, title = {Artificial Intelligence and ethics}, series = {Faster than the Future}, booktitle = {Faster than the Future}, publisher = {Digital Future Society}, address = {Barcelona}, publisher = {Hertie School}, pages = {57 -- 73}, language = {en} } @article{HaatajaBryson, author = {Haataja, Meeri and Bryson, Joanna}, title = {Reflections on the EU's AI Act and How We Could Make It Even Better}, series = {TechREG™ Chronicle}, journal = {TechREG™ Chronicle}, number = {March 2022}, editor = {Sadden, Samuel and Leyden, Andrew}, abstract = {Jurisdictions around the world are preparing regulations for artificial intelligence, as investments in AI technologies continue to increase as a source of efficiency and innovation for companies and governments. One of the most influential regulative proposals for AI is that proposed by the European Commission in April 2021, the "AI Act." The EU's proposed regulation has already inspired some international regulative proposals and is likely to broadly impact AI policies around the world. Yet the Act is still in process, it's strengths could be compromised, or it's weaknesses addressed. In this piece, we analyze the core policy concepts of the AI Act, with focus both on those worth amending and defending. These discussions may provide valuable elements for other regions beyond the EU to consider for their own AI policy. While the AI Act could still be improved to make it even more robust in managing AI-related risks to health, safety, and fundamental rights, and to increase incentives to industry to take actions beneficial to both itself and others, overall we applaud this act.}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {Europe Is in Danger of Using the Wrong Definition of AI}, series = {Wired}, journal = {Wired}, abstract = {Some intelligent systems are at risk of being excluded from oversight in the EU's proposed legislation. This is bad for both businesses and citizens.}, language = {en} } @incollection{DempseyMcBrideHaatajaetal., author = {Dempsey, Mark and McBride, Keegan and Haataja, Meeri and Bryson, Joanna}, title = {Transnational Digital Governance and Its Impact on Artificial Intelligence}, series = {The Oxford Handbook of AI Governance}, booktitle = {The Oxford Handbook of AI Governance}, publisher = {Oxford University Press}, address = {Oxford}, isbn = {9780197579329}, doi = {10.1093/oxfordhb/9780197579329.013.16}, publisher = {Hertie School}, abstract = {The rapid pace of technological advancement and innovation has put existing governance and regulatory mechanisms to the test. There is a clear need for new and innovative regulatory mechanisms that enable governments to successfully manage the integration of digital technologies into our societies, and to ensure that such integration occurs in a sustainable, beneficial, and just manner. Artificial Intelligence (AI) stands out as one of the most debated of such innovations. What exactly is it, how should it be built and deployed, how can it be used, and how should it be regulated? Yet across the period of this debate, AI is becoming widely used and addressed within existing, evolving, and bespoke regulatory contexts. The present chapter explores the extant governance of AI and, in particular, what is arguably the most successful AI regulatory approach to date, that of the European Union. The chapter explores core definitional concepts, shared understandings, values, and approaches currently in play. It argues that not only are the Union's regulations locally effective, but, due to the so-called "Brussels effect," regulatory initiatives within the European Union also have a much broader global impact. As such, they warrant close consideration.}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {Robot, all too human}, series = {XRDS: Crossroads, The ACM Magazine for Students}, volume = {25}, journal = {XRDS: Crossroads, The ACM Magazine for Students}, number = {3}, doi = {10.1145/3313131}, pages = {56 -- 59}, abstract = {Advanced robotics and artificial intelligence systems present a new challenge to human identity.}, language = {en} } @techreport{WangRichardsDorusetal., type = {Working Paper}, author = {Wang, Yifei and Richards, Marios and Dorus, Steve and Priest, Nicholas and Bryson, Joanna}, title = {Compensatory mutation can drive gene regulatory network evolution}, series = {bioRxiv}, journal = {bioRxiv}, doi = {10.1101/2019.12.18.881276}, abstract = {Gene regulatory networks underlie every aspect of life; better understanding their assembly would better our understanding of evolution more generally. For example, evolutionary theory typically assumed that low-fitness intermediary pathways are not a significant factor in evolution, yet there is substantial empirical evidence of compensatory mutation. Here we revise theoretical assumptions to explore the possibility that compensatory mutation may drive rapid evolutionary recovery. Using a well-established in silico model of gene regulatory networks, we show that assuming only that deleterious mutations are not fatal, compensatory mutation is surprisingly frequent. Further, we find that it entails biases that drive the evolution of regulatory pathways. In our simulations, we find compensatory mutation to be common during periods of relaxed selection, with 8-15\% of degraded networks having regulatory function restored by a single randomly-generated additional mutation. Though this process reduces average robustness, proportionally higher robustness is found in networks where compensatory mutations occur close to the deleterious mutation site, or where the compensatory mutation results in a large regulatory effect size. This location- and size-specific robustness systematically biases which networks are purged by selection for network stability, producing emergent changes to the population of regulatory networks. We show that over time, large-effect and co-located mutations accumulate, assuming only that episodes of relaxed selection occur, even very rarely. This accumulation results in an increase in regulatory complexity. Our findings help explain a process by which large-effect mutations structure complex regulatory networks, and may account for the speed and pervasiveness of observed occurrence of compensatory mutation, for example in the context of antibiotic resistance, which we discuss. If sustained by in vitro experiments, these results promise a significant breakthrough in the understanding of evolutionary and regulatory processes.}, language = {en} } @article{StewartMcCartyBryson, author = {Stewart, Alexander J and McCarty, Nolan and Bryson, Joanna}, title = {Polarization under rising inequality and economic decline}, series = {arXiv}, journal = {arXiv}, number = {807.11477v2}, abstract = {Social and political polarization is a significant source of conflict and poor governance in many societies. Thus, understanding its causes has become a priority of scholars across many disciplines. Here we demonstrate that shifts in socialization strategies analogous to political polarization and identity politics can arise as a locally-beneficial response to both rising wealth inequality and economic decline. Adopting a perspective of cultural evolution, we develop a framework to study the emergence of polarization under shifting economic environments. In many contexts, interacting with diverse out-groups confers benefits from innovation and exploration greater than those that arise from interacting exclusively with a homogeneous in-group. However, when the economic environment favors risk-aversion, a strategy of seeking low-risk interactions can be important to maintaining individual solvency. To capture this dynamic, we assume that in-group interactions have a lower expected outcome, but a more certain one. Thus in-group interactions are less risky than out-group interactions. Our model shows that under conditions of economic decline or increasing wealth inequality, some members of the population benefit from adopting a risk-averse, in-group favoring strategy. Moreover, we show that such in-group polarization can spread rapidly to the whole population and persist even when the conditions that produced it have reversed. Finally we offer empirical support for the role of income inequality as a driver of affective polarization in the United States, mirroring findings on a panel of developed democracies. Our work provides a framework for studying how disparate forces interplay, via cultural evolution, to shape patterns of identity, and unifies what are often seen as conflicting explanations for political polarization: identity threat versus economic anxiety.}, language = {en} } @incollection{BrysonBogani, author = {Bryson, Joanna and Bogani, Ronny}, title = {Robot Nannies Will Not Love}, series = {The Love Makers}, booktitle = {The Love Makers}, editor = {Campbell, Aifric}, publisher = {Goldsmiths Press}, address = {London}, isbn = {97819126858442}, publisher = {Hertie School}, pages = {249 -- 258}, abstract = {How artificial intelligence and robotics are transforming the future of love and desire: a philosophical thriller and essays.A chance encounter between two women and a road trip into the future: It's Christmas Eve, and Scarlett, banker-turned-technologist, is leaving a secret underground lab to catch the last flight that will get her home in time to open presents with her three-year-old son. She offers a lift to a young woman in distress, who shares her intimate life story as they drive to the airport. These revelations will have devastating consequences for both of them. The Love Makers is a philosophical thriller about female friendship, class, motherhood, women, and work--and how artificial intelligence and robotics are transforming the future of love and desire. Aifric Campbell combines her novel with essays from leading scientists and commentators who examine what's at stake in our human-machine relationships. What is our future as friends, parents, lovers? Will advances in intelligent machines reverse decades of progress for women? From robot nannies to generative art and our ancient dreams of intelligent machines, The Love Makers blends storytelling with science communication to investigate the challenges and opportunities of emergent technologies and how we want to live. ContributorsRonny Bogani, Joanna J. Bryson, Julie Carpenter, Stephen Cave, Anita Chandran, Peter R. N. Childs, Kate Devlin, Kanta Dihal, Mary Flanagan, Margaret Rhee, Amanda Sharkey, Roberto Trotta, E. R. Truitt, and Richard Watson}, language = {en} } @incollection{Bryson, author = {Bryson, Joanna}, title = {Never Love a Robot: Romantic Companions and the Principle of Transparency}, series = {The Love Makers}, booktitle = {The Love Makers}, editor = {Campbell, Aifric}, publisher = {Goldsmiths Press}, address = {London}, isbn = {97819126858442}, publisher = {Hertie School}, pages = {283 -- 290}, abstract = {It's Christmas Eve when Scarlett leaves an underground tech lab to catch the last flight that will get her home in time to open presents with her three year-old son. She offers a lift to a young woman in distress, who shares her intimate life story as they drive to the airport, but these revelations will have devastating consequences for both of their lives. Aifric Campbell combines her novel with essays from leading scientists and commentators who examine what's at stake in our human-machine relationships. What is our future as friends, parents, lovers? What is the future of love? Will advances in intelligent machines reverse decades of progress for women? From robot nannies to generative art and our ancient dreams of intelligent machines, The Love Makers blends storytelling with science communication to investigate the challenges and opportunities of emergent technologies and how we want to live.}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {The (Most) Algorithmic Animal. Unknowable Causal Structures in the Information Age}, series = {Journal for the Cognitive Science of Religion}, volume = {8}, journal = {Journal for the Cognitive Science of Religion}, number = {2}, doi = {10.1558/jcsr.23612}, pages = {115 -- 121}, abstract = {Rituals are a means of regulation - they are a means for maintaining coherence and attaining long-term goals, including social coherence. But does their efficacy depend entirely, or at all, on their opacity? In this requested commentary on Harvey Whitehouse's new book, The Ritual Animal, I discuss the utility of costly rituals in an evolutionary context, and suggest that causal opacity is only one, potentially substitutable cost. I relate this to the urgent topical concerns of polarization and of regulating sustainability globally.}, language = {en} } @article{RacineBryson, author = {Racine, Elise and Bryson, Joanna}, title = {Epidemic modeling as a means to reimagine health education and policy post-COVID}, series = {Health Education}, volume = {ahead of print}, journal = {Health Education}, issn = {0965-4283}, doi = {10.1108/HE-02-2021-0028}, abstract = {Purpose As illustrated by coronavirus disease 2019 (COVID-19), epidemic models are powerful health policy tools critical for disease prevention and control, i.e. if they are fit for purpose. How do people ensure this is the case and where does health education fit in? Design/methodology/approach This research takes a multidisciplinary approach combining qualitative secondary and primary data from a literature review, interviews and surveys. The former spans academic literature, grey literature and course curriculum, while the latter two involve discussions with various modeling stakeholders (educators, academics, students, modeling experts and policymakers) both within and outside the field of epidemiology. Findings More established approaches (compartmental models) appear to be favored over emerging techniques, like agent-based models. This study delves into how formal and informal education opportunities may be driving this preference. Drawing from other fields, the authors consider how this can be addressed. Practical implications This study offers concrete recommendations (course design routed in active learning pedagogies) as to how health education and, by extension, policy can be reimagined post-COVID to make better use of the full range of epidemic modeling methods available. Originality/value There is a lack of research exploring how these methods are taught and how this instruction influences which methods are employed. To fill this gap, this research uniquely engages with modeling stakeholders and bridges disciplinary silos to build complimentary knowledge.}, language = {en} } @incollection{McBrideKupiBryson, author = {McBride, Keegan and Kupi, Maximilian and Bryson, Joanna}, title = {Untangling Agile Government: On the Dual Necessities of Structure and Agility}, series = {Agile Government: Emerging Perspectives in Public Management}, booktitle = {Agile Government: Emerging Perspectives in Public Management}, editor = {Stephens, Melodena and Awamleh, Raed and Salem, Fadi}, publisher = {World Scientific Publishing}, address = {Singapur}, isbn = {9789811239694}, doi = {10.1142/9789811239700_0002}, publisher = {Hertie School}, pages = {300}, abstract = {The governments of today are not able to transform and adapt to changes in the world around them, as demanded by their constituents. The nature of work, value of public goods, and the constant bombardment of crises are making the old bureaucratic structures obsolete. Agile Government is an emerging theme, that of government-wide reinvention for adaptiveness and responsiveness. It places the accountability, delivery, capture, design and creation of public value at the heart of the government. The concept of agile government is confused with terms like Agile Manifesto, agile governance, agility among others, and because of this, needs some unpacking. This book is a deep dive into this topic. It offers insights from the theoretical development of the topic of agile government, some lessons from government practices around the world, and ongoing academic and policy research. The project is spearheaded by the Mohammed Bin Rashid School of Government, which is the first teaching and research institution in the Arab world focusing on public policy and governance.}, language = {en} } @unpublished{DempseyMcBrideHaatajaetal., author = {Dempsey, Mark and McBride, Keegan and Haataja, Meeri and Bryson, Joanna}, title = {Transnational digital governance and its impact on artificial intelligence}, doi = {10.31235/osf.io/xu3jr}, pages = {20}, abstract = {The rapid pace of technological advancement and innovation has put governance and regulatory mechanisms to the test. There is a clear need for new and innovative regulatory mechanisms that enable governments to successfully manage the integration of such technologies into our societies and ensure that such integration occurs in a sustainable, beneficial, and just manner. Artificial Intelligence stands out as one of the most debated such innovations. What exactly is it, how should it be built, how can it be used, and how and should it be regulated? Yet, in this debate, AI is becoming widely utilized within both existing, evolving, and bespoke regulatory contexts. The present chapter explores in particular what is arguably the most successful AI regulatory approach to date, that of the European Union. We explore core definitional concepts, shared understandings, values, and approaches currently in play. We argue that due to the so-called 'Brussels effect', regulatory initiatives within the European Union have a much broader global impact and, therefore, warrant close inspection.}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {One Day, AI Will Seem as Human as Anyone. What Then?}, series = {Wired}, journal = {Wired}, language = {en} } @article{BrysonHaataja, author = {Bryson, Joanna and Haataja, Meeri}, title = {The European Parliament's AI Regulation: Should We Call It Progress?}, series = {Amicus Curiae}, volume = {4, Series 2}, journal = {Amicus Curiae}, number = {3}, doi = {10.14296/ac.v4i3.5612}, pages = {707 -- 718}, abstract = {The European Union (EU) has been leading the world with its influential digital regulation. However, the EU's legislative process is sufficiently complex and careful that some national legislation clearly influenced by the EU's AI Regulation is already in place in other countries, before the law has even been finalized in the EU. Meanwhile, other states and regions are just beginning to develop AI policy. For both the EU and such others, we here describe the outcomes of the first round of legislative action by one of the EU's two legislative bodies, the European Parliament, in terms of modifying the Artificial Intelligence Act. The Parliament has introduced a number of changes we consider to be enormously important, some in a very good way, and some in a very bad way. At stake is whether the AI Act really brings the power and strength of product law to continuously scale improved practice on products in the EU with intelligent components, or whether the law becomes window-dressing aimed only at attacking a few elite actors post hoc. We describe here the EU process, the changes and our recommendations.}, language = {en} } @techreport{DariusStockmannBrysonetal., type = {Working Paper}, author = {Darius, Philipp and Stockmann, Daniela and Bryson, Joanna and Cingolani, Luciana and Griffin, Rachel and Hammerschmid, Gerhard and Kupi, Maximilian and Mones, Haytham and Munzert, Simon and Riordan, R{\´o}n{\´a}n and Stockreiter, Simona}, title = {Implementing Data Access of the Digital Services Act: Collaboration of European Digital Service Coordinators and Researchers in Building Strong Oversight over Social Media Platforms}, doi = {10.48462/opus4-4947}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:b1570-opus4-49479}, pages = {11}, abstract = {The EU Digital Service Acts signals a move away from self-regulation towards co-regulation of social media platforms within the European Union. To address online harms and rising platform power the DSA clarifies responsibilities of platforms and outlines a new technology regulatory framework to increase oversight. One key oversight instrument constitutes Article 40 of the DSA, which lays out data access for vetted researchers, who add value to regulators and the broader public as creators of knowledge, educators, advisors, innovators, and watchdogs. Currently, the EU Commission and national governments make important decisions regarding Digital Service Coordinators (DSCs) that play a key role in implementation. Based on expertise on European public administration and political science we lay out key challenges and success factors of DSCs that will play a role in promoting successful cooperation between DSCs and researchers. We provide three recommendations: First, we recommend to strengthen transfer of scientific knowledge into policy-making by processing publicly accessible publications within public administrative bodies. To this end, capacities of DSCs need to be increased. In addition, we also point towards the database of vetted researchers collected by the Board of DSCs as important resource in order to strengthen knowledge transfer. Second, the DSC network requires agile institutions with fast response time in order to enable researchers to play a constructive role in implementation. This also includes institutional procedures between DSCs and the Intermediary Body and Data Protection Agencies. To avoid delay in implementation agile institution-building needs to start now. Finally, institutional safeguards will help to avoid strategic choice of companies of the DSC of establishment. At the same time, the Irish DSC's capacity should be strengthened compared to other national DSCs since most large intermediary services providers have their European headquarters in Ireland.}, language = {en} } @article{BaumBrysonDignumetal., author = {Baum, Kevin and Bryson, Joanna and Dignum, Frank and Dignum, Virginia and Grobelnik, Marko and Hoos, Holger and Irgens, Morten and Lukowicz, Paul and Muller, Catelijne and Rossi, Francesca and Shawe-Taylor, John and Theodorou, Andreas and Vinuesa, Ricardo}, title = {From Fear to Action: AI Governance and Opportunities for All}, series = {Frontiers in Computer Science}, volume = {5}, journal = {Frontiers in Computer Science}, doi = {10.3389/fcomp.2023.1210421}, language = {en} } @article{BackovskyBryson, author = {Backovsky, David and Bryson, Joanna}, title = {Going Nuclear? Precedents and Options for the Transnational Governance of AI}, series = {Horizons}, journal = {Horizons}, number = {Summer 2023, No.24}, issn = {2406-0402}, pages = {84 -- 95}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {Margaret Boden obituary: cognitive scientist who explored how machines might emulate human imagination}, series = {Nature}, volume = {644}, journal = {Nature}, number = {8077}, publisher = {Springer Science and Business Media LLC}, doi = {10.1038/d41586-025-02548-0}, pages = {603 -- 603}, language = {en} } @article{Bryson, author = {Bryson, Joanna}, title = {Human Experience and AI Regulation: What European Union Law Brings to Digital Technology Ethics}, series = {Weizenbaum Journal of the Digital Society}, volume = {3}, journal = {Weizenbaum Journal of the Digital Society}, number = {3}, issn = {2748-5625}, doi = {10.34669/WI.WJDS/3.3.8}, abstract = {Although nearly all artificial intelligence (AI) regulatory documents now reference the importance of human-centering digital systems, we frequently see AI ethics itself reduced to limited concerns, such as bias and, sometimes, power consumption. Although their impacts on human lives and our ecosystem render both of these absolutely critical, the ethical and regulatory challenges and obligations relating to AI do not stop there. Joseph Weizenbaum described the potential abuse of intelligent systems to make inhuman cruelty and acts of war more emotionally accessible to human operators. But more than this, he highlighted the need to solve the social issues that facilitate violent acts of war, and the immense potential the use of computers offers in this context. The present article reviews how the EU's digital regulatory legislation—well enforced—could help us address such concerns. I begin by reviewing why the EU leads in this area, considering the legitimacy of its actions both regionally and globally. I then review the legislation already protecting us—the General Data Protection Regulation, the Digital Services Act, and the Digital Markets Act—and consider their roles in achieving Weizenbaum's goals. Finally, I consider the almost-promulgated AI Act before concluding with a brief discussion of the potential for future enforcement and global regulatory cooperation.}, language = {en} } @inproceedings{MellmannArbuzovaKontogiorgosetal., author = {Mellmann, Heinrich and Arbuzova, Polina and Kontogiorgos, Dimosthenis and Yordanova, Magdalena and Haensel, Jennifer X. and Hafner, Verena V. and Bryson, Joanna}, title = {Effects of Transparency in Humanoid Robots - A Pilot Study}, series = {Companion of the 2024 ACM/IEEE International Conference on Human-Robot Interaction}, booktitle = {Companion of the 2024 ACM/IEEE International Conference on Human-Robot Interaction}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, isbn = {9798400703232}, doi = {10.1145/3610978.3640613}, pages = {750 -- 754}, abstract = {Transparency is recognized as a vital feature for understanding and predicting robot behavior. Another feature that affects interaction with robots is their anthropomorphism. The relationship between these remains under-explored but is postulated to be negative. We present a pilot study investigating the effects of robot transparency in human-robot interactions, where the robot has an anthropomorphic appearance. We asked participants to evaluate and interact with the humanoid robot Pepper to examine whether visualizing the robot's goals and behavior affects perceived intelligence, anthropomorphism, and robot agency. Our preliminary findings suggest that users may attribute higher ratings of agency when interacting with a robot visualizing its goals. In this late-breaking report, we propose our experiment on the interplay between transparency and anthropomorphism in human-robot interaction and summarize insights from our preliminary pilot study.}, language = {en} } @misc{Bryson, author = {Bryson, Joanna}, title = {Governing society with flexible AI}, series = {Science}, volume = {383}, journal = {Science}, number = {6688}, issn = {0036-8075}, doi = {10.1126/science.adn6814}, pages = {1185 -- 1185}, abstract = {Policy that prioritizes human agency is possible, even in an age of artificial intelligence}, language = {en} } @techreport{SchmitzBryson, type = {Working Paper}, author = {Schmitz, Chris and Bryson, Joanna}, title = {A Moral Agency Framework for Legitimate Integration of AI in Bureaucracies}, number = {v3}, publisher = {arXiv}, doi = {10.48550/arXiv.2508.08231}, pages = {15}, abstract = {Public-sector bureaucracies seek to reap the benefits of artificial intelligence (AI), but face important concerns about accountability and transparency when using AI systems. In particular, perception or actuality of AI agency might create ethics sinks — constructs that facilitate dissipation of responsibility when AI systems of disputed moral status interface with bureaucratic structures. Here, we reject the notion that ethics sinks are a necessary consequence of introducing AI systems into bureaucracies. Rather, where they appear, they are the product of structural design decisions across both the technology and the institution deploying it. We support this claim via a systematic application of conceptions of moral agency in AI ethics to Weberian bureaucracy. We establish that it is both desirable and feasible to render AI systems as tools for the generation of organizational transparency and legibility, which continue the processes of Weberian rationalization initiated by previous waves of digitalization. We present a three-point Moral Agency Framework for legitimate integration of AI in bureaucratic structures: (a) maintain clear and just human lines of accountability, (b) ensure humans whose work is augmented by AI systems can verify the systems are functioning correctly, and (c) introduce AI only where it doesn't inhibit the capacity of bureaucracies towards either of their twin aims of legitimacy and stewardship. We suggest that AI introduced within this framework can not only improve efficiency and productivity while avoiding ethics sinks, but also improve the transparency and even the legitimacy of a bureaucracy.}, language = {en} } @incollection{Bryson, author = {Bryson, Joanna}, title = {From Definition to Regulation: Is the European Union Getting AI Right?}, series = {Artificial Intelligence and Fundamental Rights: The AI Act of the European Union and its implications for global technology regulation}, volume = {4}, booktitle = {Artificial Intelligence and Fundamental Rights: The AI Act of the European Union and its implications for global technology regulation}, editor = {Raue, Benjamin and von Ungern-Sternberg, Antje and Kumkar, Lea Katharina and R{\"u}fner, Thomas}, publisher = {Verein f{\"u}r Recht und Digitalisierung e.V. Institute for Digital LawTrier (IRDT)}, address = {Trier}, isbn = {9783565013197}, doi = {10.25353/ubtr-dab1-9b5c-1ec6}, publisher = {Hertie School}, pages = {11 -- 34}, language = {en} } @article{EvansRobbinsBryson, author = {Evans, Katie D. and Robbins, Scott A. and Bryson, Joanna}, title = {Do We Collaborate With What We Design?}, series = {Topics in Cognitive Science}, volume = {17}, journal = {Topics in Cognitive Science}, number = {2}, doi = {10.1111/tops.12682}, pages = {392 -- 411}, abstract = {The use of terms like "collaboration" and "co-workers" to describe interactions between human beings and certain artificial intelligence (AI) systems has gained significant traction in recent years. Yet, it remains an open question whether such anthropomorphic metaphors provide either a fertile or even a purely innocuous lens through which to conceptualize designed commercial products. Rather, a respect for human dignity and the principle of transparency may require us to draw a sharp distinction between real and faux peers. At the heart of the concept of collaboration lies the assumption that the collaborating parties are (or behave as if they are) of similar status: two agents capable of comparable forms of intentional action, moral agency, or moral responsibility. In application to current AI systems, this not only seems to fail ontologically but also from a socio-political perspective. AI in the workplace is primarily an extension of capital, not of labor, and the AI "co-workers" of most individuals will likely be owned and operated by their employer. In this paper, we critically assess both the accuracy and desirability of using the term "collaboration" to describe interactions between humans and AI systems. We begin by proposing an alternative ontology of human-machine interaction, one which features not two equivalently autonomous agents, but rather one machine that exists in a relationship of heteronomy to one or more human agents. In this sense, while the machine may have a significant degree of independence concerning the means by which it achieves its ends, the ends themselves are always chosen by at least one human agent, whose interests may differ from those of the individuals interacting with the machine. We finally consider the motivations and risks inherent to the continued use of the term "collaboration," exploring its strained relation to the concept of transparency, and consequences for the future of work.}, language = {en} } @article{JuglPagelGarciaJimenezetal., author = {Jugl, Marlene and Pagel, William A. M. and Garcia Jimenez, Maria Camilla and Salendres, Jean Pierre and Lowe, Will and Malikova, Helena and Bryson, Joanna}, title = {Spamming the regulator: exploring a new lobbying strategy in EU competition procedures}, series = {Journal of Antitrust Enforcement}, volume = {12}, journal = {Journal of Antitrust Enforcement}, number = {1}, doi = {10.1093/jaenfo/jnad009}, abstract = {Regulation plays a central role in modern governance; yet, we have limited knowledge of how subjects of regulation—particularly, private actors—act in the face of potentially adverse regulatory decisions. Here, we document and examine a novel lobbying strategy in the context of competition regulation, a strategy that exploits the regulator's finite administrative capacities. Companies with merger cases under scrutiny by the European Commission's Directorate General for Competition appear to be employing a strategy of 'spamming the regulator,' through the strategic and cumulative submission of economic expert assessments. Procedural pressures may result in an undeservedly favourable assessment of the merger. Based on quantitative and qualitative analyses of an original dataset of all complex merger cases in the European Union 2005-2020, we present evidence of this new strategy and a possible learning process among private actors. We suggest remedies to ensure regulatory effectiveness in the face of this novel strategy.}, language = {en} } @techreport{BaumBryson, type = {Working Paper}, author = {Baum, Leonard and Bryson, Joanna J.}, title = {Policy lessons from China: A quantitative examination of China's new competition regime for the digital economy}, publisher = {Center for Open Science}, doi = {10.31235/osf.io/zyc6s_v3}, pages = {36}, abstract = {Growing global concern about the problems associated with concentrated market power in the digital economy is leading to a renewed interest in competition policy. Since the late 2010s, China's government has squarely confronted the problems of its own 'Big Tech' with a new competition regime for digital markets. Outcomes represent a unique learning opportunity for Western academics, competition authorities and lawmakers alike, which has so far been underutilized. However, given unreliable official figures, a new methodology is needed to assess competition in China's digital economy. This article introduces a market capitalization approach that builds on the informativeness of China's financial markets. We use Bloomberg financial data of 1142 publicly listed firms for the period 2019 to 2022 to quantitatively examine the impact of China's new digital competition regime. We find a causal link between the new governance approach and a reduction of market concentration and aggregate growth in the primary markets of China's three most dominant digital platforms - Baidu, Alibaba and Tencent (BATs). Further, our results show a robust correlation between the new competition regime and reduced market concentration and market capitalization growth rates across China's digital markets. Other empirical findings include a negative correlation between market concentration and the openness of digital markets, a non-relationship between market concentration and profits, and the inability of profit and revenue-based metrics to capture market power effectively in China's digital economy. Finally, we discuss the relevance of these insights for Western regulatory strategies, particularly as the EU and China emerge as global frontrunners in the field of digital competition regulation.}, language = {en} } @techreport{HeddesheimerBryson, type = {Working Paper}, author = {Heddesheimer, Vincent and Bryson, Joanna J.}, title = {Economic Insecurity Increases Affective Polarization and Outgroup-Aversion}, publisher = {Center for Open Science}, doi = {10.31219/osf.io/x23bj_v1}, pages = {19, S21}, abstract = {Affective polarization --- a deep emotional divide between opposing social and political groups --- has become a pressing global governance challenge, disrupting both democratic processes and policy uptake. While some degree of position polarization should be expected in a healthy, multi-party political system, the factors driving recent extremes in divisions remain unclear. One candidate factor, economic instability, has been frequently linked to rising polarization, but causal evidence to date has been limited. Using panel survey data from the Netherlands and a difference-in-differences design, we show that economic insecurity significantly increases affective polarization. The effect persists for years, with a longer (though slower) expression among men. It is also accompanied by increased distrust and social group clustering. Our findings suggest that economic insecurity is a key driver of polarization, highlighting the role of labor market stability in shaping political and social cohesion.}, language = {en} }