An essay for the McQuillian Institute’s project: ‘A Constitution for the Twenty-First Century’

By Mike Hulme

Summary

The western liberal democratic tradition has evolved to place a very high value on science and its relevance, if not necessity, for public policy decision-making. A powerful exemplar of this view over recent decades has been the design, operation and reception of the UN’s Intergovernmental Panel on Climate Change (IPCC). On the one hand, the IPCC has been enormously successful, generating ‘IPCC-envy’ among observers and activists representing a wide range of other non-climatic grand challenges. And yet the ‘wicked’ nature of the climate change problem has also exposed the limits of scientific knowledge with respect to policymaking and, more generally, to the healthy operation of democracy. This essay explores what can be learned from the ‘IPCC-case’ with respect to the relationship between science (and expertise more generally), democracy and policy with respect to many of the pressing transnational challenges societies will face in the future.

The western liberal democratic tradition has evolved to place a very high value on science and its relevance, if not its necessity, for public policy decision-making. ‘Evidence-based policy’ is a slogan that, for many, has becomes a touchstone of public rationality. Yet there is a tension in this view of science that reveals itself in various ways. In recent decades this tension has turned into a worrying problem for democracies. Science claims for itself the ability to generate reliable forms of specialised knowledge that are not discoverable by ordinary people, by the demos.[1] Scientific expertise, and hence scientific experts, has therefore gained a status in societies not available to ordinary people. The tension arises because, in its most egalitarian manifestation, a democracy gives the demos recognition, a voice and, at least some, influence on important matters of the day.

This tension between ‘expert knowledge’ (science) and the ‘voice of the people’ (the demos) has been supremely illustrated in recent decades by the phenomenon of climate change—both what is known about it and what should be done about it. Central to attempts at reconciling science and democracy in relation to this phenomenon has been the design, operation and reception of the UN’s Intergovernmental Panel on Climate Change (IPCC).

In this essay, I explore what can be learned from the ‘IPCC-case’ about the relationship between science (and scientific expertise more generally), democracy and public policy. I suggest that climate change is a limiting case for what science as a form of knowledge can offer democracies with respect to many of the pressing transnational challenges facing humankind.

The IPCC was created in 1988 as an inter-governmental institution operating under the umbrella of the United Nations. It is now into its seventh knowledge assessment cycle and in many respects it has been enormously successful—it appears to be an established part of the international climate regime, part of what might be called the constitutional settlement for climate governance. The IPCC has achieved this by securing a hegemony over climate change knowledge, its epistemic authority being recognised and valued by a wide range of political actors. There may be some interests—from both the Left and Right of the political spectrum—that seek to undermine its authority. But no-one who wants to speak meaningfully about climate policy can ignore the IPCC, either to defer to it or to discredit it.

These successes have generated what I and some others have called ‘IPCC-envy’ among issue entrepreneurs and policy activists representing a wide range of other non-climatic grand challenges.[2] Initially, this ‘envy’ emerged with respect to other recognisably ‘environmental’ topics, such as biodiversity, food systems, chemical pollution and waste, soil management, and desertification. But in recent years, this ‘envy’ has manifest itself in relation to issues such as Artificial Intelligence governance, migration and asylum, sustainable growth and, most recently, global inequality.[3]

And yet the ‘wicked’ nature of climate change has exposed what I believe to be the limits of scientific knowledge with respect to transnational negotiation, cooperation, and policymaking and, more generally, with respect to the healthy operation of democracies.[4] There are three (related) reasons why these limits have been exposed in the case of climate change.

First, over the past few decades, and certainly since the IPCC formed in 1988, the nature of science itself has changed. No longer is science the self-governing autonomous ‘republic’ imagined by Michael Polanyi in his widely-read 1962 essay. Science can only defend its social contract by allowing the demos to share in its internal governance, at least to some degree.[5] This is hard for science to do. Second, it has become increasingly clear, even if it wasn’t so in the 1980s, that the underlying drivers, impacts and responses to climate change connect with nearly all human activities and institutions in some way—whether spatially, temporally, topically, or socially, whether expressed through economics, technology or human values. The politics of climate change have therefore become intractable and well beyond the ability of scientific knowledge to resolve. Witness, for example, the diminishing returns from the annual Conference of the Parties to the UN’s Framework Convention on Climate Change, despite the authoritative reports of the IPCC.

The third reason why the limits of knowledge for governing climate have been exposed is the greatly raised expectations of the demos to be given voice in public affairs, at least in some of the roughly half of countries retaining the vestiges of democracy. Expertise is less deferred to than it once was, and publics are generally more suspicious of science. These changing expectations have been fuelled by the proliferation of new digital communication platforms—the internet, social media and now AI (to which I will return)—which propagate the voice of the demos rapidly and easily.

For at least these three reasons, the old model of the IPCC—offering the world the ‘guiding hand of science’ for tackling the many layered political challenges involved in defusing climate risks—no longer holds the promise it seemingly once did.[6] The recent travails of the IPCC’s Seventh Assessment Report, taking place against a background of rising geopolitical tension, diminishing faith in international institutions, and political populism, bears witness to this.[7]

What responses to this situation can be discerned and how do these responses change the relationship between science and democracy? I suggest three.

First is the response of the IPCC itself—resulting from a range of external pressures—which is to try and diversify the knowledge base that it recognises and assesses. For example, earlier this year the IPCC held an expert workshop in the UK to discuss “engaging diverse knowledge systems”, with the intention “to consider what systems of knowledge — including scientific, Indigenous Knowledge systems, local knowledge systems — can be accessed and assessed by the IPCC within the framework of existing principles and procedures.”[8]

However, this response exacerbates a further problem the IPCC faces, namely, maintaining a clear distinction between facts and values, a long-standing principle of the IPCC expressed through its slogan, ‘policy relevant but policy neutral’. So, a second response to the challenges the IPCC faces has been the proposal from some commentators for it to abandon such boundary work and embrace the inevitability that when faced with a wicked problem like climate change such value neutrality cannot be maintained. For example, this pathway has been articulated by Noel Castree and colleagues, who call for a new era of global environment assessments (GEAs), ones they term “GEA 3.0”.[9] Such assessment processes would explicitly “introduce political reasoning and structured normative debate about existential alternatives, a pre-requisite to strategic decision-making and its operational expression.”

I have some sympathy with this view. In an earlier essay, I argued that “investing in new participatory and agonistic forms of democracy—where value conflicts and political disagreements are acknowledged, voiced, and worked with—is as important (perhaps more important) than investing in new scientific or technical knowledge.[10] I am less optimistic than Castree, however, that this goal could be operationalised in the case of climate change, or indeed any other transnational environmental challenge, through a re-structuring of the IPCC.

A third response to the IPCC’s dilemma is more radical but will in due course be necessary: how to harness the emerging capabilities of AI. Holly Buck and colleagues, among others, have begun to sketch out what some of these options might look like. They offer four difference scenarios which they label ‘the right of agents’, ‘superior truth machine’, ‘anticipatory resistance’, and ‘public backlash’.[11] As they and others point out, there is a huge problem here are delegating human judgements, let alone democratic deliberation, to machines. Already we are seeing how Large Language Models are able to forecast the outcome of large social-science experiments, an outcome which subordinates human preferences and behaviour to an algorithm.[12] Machen and Pearce summarise the danger in clear terms: “The implementation of AI may run counter to norms of democratic accountability. These challenges represent an urgent dilemma for climate governance as ignoring these issues will erode democratic oversight, lead to unpopular unintended consequences, and could reverse recent positive trends in diversity and participation within climate science and policy.”[13]

Each of these imagined responses by the IPCC keeps a core assumption intact: that knowledge, however formulated, is the essential asset for guiding the world into a future where the dangers of a changing climate are lessened. It is a questionable assumption that knowledge has ever been able, purposefully, to guide the world’s future, as the case of the IPCC has shown. The rise of AI will make this assumption even more tenuous as the technology insinuates its way into all aspects of human life, including our ability for independent judgement and our sense of self.

 In his 2008 book ‘Climate Wars: The Fight for Survival as the World Overheats’, security expert Gwynne Dyer developed a series of speculative scenarios about the dire consequences of climate change for the future of the world. However, he also speculated that “by the 2020s we may be plunged into a struggle over the proper role of artificial intelligence that is just as important to the future of the human race as [is] getting our impact on the climate under control.”[14]  We are nearing the end of the 2020s and, sure enough, there is a lively “struggle” over the “proper role” for AI. My fear is that while it may expand our “knowledge”, it will do so at the cost of making climate even less governable than it is already.

Copyright, Mike Hulme, University of Cambridge, August 2026

[1] For a discussion of this, see: Collins, H. (2014) Are We All Scientific Experts Now? Cambridge: Polity.

[2] Hulme, M. (2024) Why do other science-policy issues want an IPCC, and should they? Dialogues on Climate Change. 1(1): 7-11; De Dona, M. (2025) “IPCC?envy”? Shaping global soil and land governance through science?policy activism. Environment, Development and Sustainability. 27: 12,827-844.

[3] Anon (2026) Why there needs to be a global debate on inequality. Nature. 655(17 July): 547.

[4] Of course, less than half of the 194 countries signed up to the Paris Agreement are democracies, according to the Economist Intelligence Unit’s Democracy Index (of the 167 tracked nations, slightly less than half are classified as either “full democracies” or “flawed democracies”).

[5] Polanyi, M. (1962) The republic of science: its political and economic theory. Minerva. 1: 54-74; Tyler, C., Akerlof, K.L. et al. (2026) Six ways to put the public at the heart of science and policy. Nature. 655: 34-37.

[6] See Granado, C. (2025) Scientific expertise in early international negotiations on climate change: Bert Bolin and the IPCC. British Journal for the History of Science. 1–23. doi:10.1017/S0007087425101052. See also De Pryck, K. and Hulme, M. (eds.) (2022) A Critical Assessment of the IPCC. Cambridge University Press.

[7] Petersmann, M. (2026) The IPCC in the post-Paris climate regime – locked into institutional inertia? Earth System Governance. 29: 100349.

[8] See: https://www.ipcc.ch/event/ipcc-workshop-on-engaging-diverse-knowledge-systems-and-ipcc-workshop-on-methods-of-assessment/. See also: Van Bavel, B. et al. (2022) Indigenous knowledge systems. Chapter 13 in: De Pryck, K. & Hulme, M. (eds.) A Critical Assessment of the IPCC. Cambridge University Press.

[9] Castree, N., Bellamy, R. and Osaka, S. (2021) The future of global environmental assessments: Making a case for fundamental change. Anthropocene Review. 8(1): 56-82.

[10] Hulme, M. (2020) One Earth, many futures, no destination. One Earth. 2(4): 309-311.

[11] Buck, H-J., Al Khourdajie, A. et al. (2026) Potential futures for the IPCC’s approach to artificial intelligence. Communications Earth & Environment. 7: 372. 8pp.

[12] Ashokkumar, A. et al. (2026) Large language models can predict the results of social science experiments. Nature. 656: 115-122.

[13] Machen, R. and Pearce, W. (2025) Anticipating the challenges of AI in climate governance: An urgent dilemma for democracies. WIREs Climate Change. 16(2): e70002.

[14] p. xv in: Dyer, G. (2008) Climate Wars: The Fight for Survival as the World Overheats. Brunswick, Australia: Scribe Publications.