Oxford AgoraThe Global Essay and Defence Prize

A system may be more accurate, consistent or efficient than a human decision-maker without possessing legitimate authority of its own. Who should determine when these advantages justify delegation: the people who develop the system, the institutions that deploy it, those affected by its decisions or the wider public? Should the required standard depend on whether the decision concerns medical diagnosis, criminal justice, finance or transport? When people disagree about what a “better” decision means, can performance alone settle who should decide?

A language model may produce convincing explanations, solve unfamiliar problems and respond appropriately to context. Yet these abilities might result from statistical prediction rather than an understanding of meaning. Should understanding be inferred from behaviour, from the internal representations of a system or from its ability to form and revise models of the world? If no possible experiment could distinguish understanding from successful imitation, does the distinction remain scientifically meaningful?

Assume that an inherited trait could be changed without unintended genetic errors. Would preventing a fatal single-gene disorder be fundamentally different from reducing susceptibility to a complex disease, altering physical characteristics or influencing cognitive traits? Genetic effects also depend on other genes and on the environment, so what would it mean to alter a trait predictably? Who should define the boundary between treatment and enhancement, and could a choice that begins as voluntary become a social expectation?

Cancer is not a fixed target, but an evolving population of cells. Intensive treatment may destroy vulnerable cells while allowing resistant cells to survive, multiply and dominate the tumour. Adaptive therapy instead attempts to manage this competition, sometimes by reducing or interrupting treatment rather than administering the maximum tolerated dose. Under what biological conditions could controlling a tumour be more effective than attempting to eradicate it, and what clinical evidence would be needed to change established practice?

Nuclear power can provide continuous low-carbon electricity, but concerns remain about construction time, cost, accidents, radioactive waste and weapons proliferation. A system with little or no nuclear power would depend more heavily on renewables, storage, transmission, flexible demand and other forms of firm generation. How should familiar but potentially severe nuclear risks be compared with the less visible system-wide risks of excluding nuclear power? Should the answer differ between countries with different resources, institutions and energy needs?

Ice sheets, ocean circulation, forests and other components of the Earth system may respond gradually to pressure before shifting abruptly into a different state. Scientists have proposed early-warning signals, but similar patterns may also arise from natural variability, limited observations or weaknesses in climate models. What evidence would show that a system is approaching a genuine threshold? How should policy respond when the probability of crossing a tipping point is deeply uncertain but the consequences may be irreversible?

Public policy often gives less weight to costs and benefits that occur far in the future. This may reflect the expectation that future societies will be wealthier, the possibility of investing resources elsewhere or uncertainty about what the future will contain. But should the passage of time itself make a future benefit less valuable? Would refusing to discount the future protect generations that cannot represent themselves, or unfairly require poorer people today to make sacrifices for people who may be richer tomorrow?

Elections provide public authorisation, but they can favour wealth, party organisation, professional politicians and those most skilled at campaigning. Selection by lottery could create a legislature that more closely resembles the population, while deliberation and access to expert evidence might allow ordinary citizens to make informed decisions. Would such a body possess sufficient competence, independence and legitimacy? Should it advise elected politicians, scrutinise their decisions or exercise law-making power of its own?

Governments, courts, landowners and empires often produced extensive archives, while enslaved people, indigenous communities, workers and the poor left fewer written records. Can historians recover their experiences by reading official documents against the grain, drawing on oral testimony or examining objects, landscapes and demographic evidence? How should different forms of evidence be compared and corroborated? At what point does responsible reconstruction become speculation?

Present generations cannot be personally guilty of crimes committed before they were born. Yet they may inherit institutions, borders, wealth and political identities shaped by those crimes. Does this continuity create obligations without transferring guilt? Should remembrance lead to apology, restitution or material repair? Can a historical responsibility expire, and when might preserving memory obstruct rather than support reconciliation?