The Sorcerer's Apprentice and the Future of Knowledge

Return to Athens for a moment and consider the history of Western knowledge. Aristotle saw it as falling into three types, which were ranked in terms of a social hierarchy. The lowest form, productive knowledge, consisted of skill in making things. Aristotle thought this was fine within limits, but that undue attention to creature comforts led to ‘a life fit for beasts’. Higher was practical knowledge. It dealt with actions, particularly those that shape our social and political life. The greatest kind of knowledge was theoretical in nature. Sharing the root of ‘theos’ or god, it was pursued for its own sake.

Of course, that was a long time ago. Life was simple then, and often brutal. Technology was limited to craftwork, and ‘commodious living’ (Hobbes) was hardly possible.

The modern age has changed our ability to create comforts. The change was driven by epistemological innovation: knowledge was reimagined as being of a single type, rooted in what Descartes called a mathesis universalis, a universal scientific method based in mathematics. This was no mere theoretical knowledge: the book of nature was seen as written in the language of mathematics. Via technology, this method has given us unprecedented power over the natural world. It has allowed us to vastly increase our amenities. It has also increased our political power, for having control over nature means having control over others as well.

Shaped by Descartes and his brethren, modernity has been remarkably successful at its appointed task: knowledge has given us prodigious powers. Perhaps too much so: what was once a one-off – the existential danger represented by nuclear weapons – has become an increasingly common phenomenon. Artificial intelligence is just one of a range of new technologies that threaten our safety, our freedom, the meaning of our lives, even our very existence.

The assumptions concerning the production of knowledge that have ruled for the last 300 years, encapsulated by Kant's ‘sapere aude!’ (‘dare to know!’), need revising. With a few marginal exceptions – which are violated whenever convenient – the modern era has been a time of laissez faire knowledge production. Researchers freely create, and engineers make what they will, while society is left to grapple with the consequences. We see this dynamic today, for who is calling for AI, other than its inventors and investors, who stand to make billions?

It's time to place knowledge at the docket. Knowledge producers need to answer quid juris? – by what right do you pursue insights that can lead to destructive results?

In the fable The Sorcerer's Apprentice a broom is endowed with artificial intelligence. It is programmed to serve its master's wishes, but the powers it is endowed with run in unanticipated directions. In Goethe's version, ‘Spirits I have conjured, no longer pay me heed’. The optimists leading the charge on AI promise astonishing results from unregulated knowledge production, while also believing they will remain in control of the outcomes – or if not, that the outcomes will be benign. But programmers are apprentices, not sorcerers. They won’t be able to clean up all their messes.

1

This essay begins with an account of the value and efficacy of interdisciplinary approaches to knowledge. Interdisciplinarity has been widely offered as a remedy for what ails academic knowledge production. Interdisciplinary researchers work the space between academic knowledge and societal progress, presenting interdisciplinarity as the means for improving the relevance of academic knowledge.

Mid-twentieth century academia (and society) assumed that academic knowledge was by its very nature relevant. By the end of the twentieth century its relevance had become problematic: society had run what was in effect a 50-year experiment in funding research and concluded that much of what it was paying for was useless. Interdisciplinarity became the preferred answer for addressing this problem.

Set aside for the moment whether interdisciplinarity was successful: we now live in different times. Each of the elements of our social epistemology have changed: the nature of knowledge, the places and ways where it is produced, and the effects it has on society. We face questions that interdisciplinarity never thought to ask: is the lack of relevance the only problem with knowledge production? Does new knowledge always constitute progress, or has knowledge production increasingly led to trivialising or threatening results? Do we need a governor on knowledge production? And what do these issues suggest about the future of the university?

Knowledge production has become too powerful to continue along the laissez faire path it has trod. The concept of relevance, the stock and trade of interdisciplinarity, is now problematic. For knowledge production today is often too relevant – too distracting, dangerous, and seductive. Rather than relevance, the conversation we need about knowledge production concerns its governance.

It's time to bring the interdisciplinary project to a close. ¹ I explain this view in the first half of this essay. The second half offers a sketch of where the conversation should go next, exploring the possibility of a new social consensus on values, the necessary condition for the creation of a governance structure for knowledge. I then end with thoughts on the role that universities could play within a reconstituted politico-epistemic order.

2

Interdisciplinarity – the blending of disciplines to increase the power and relevance of academic work – still retains a bit of the au courant. This is odd, since the term dates from 1926, and the idea has been in wide circulation since the 1970s. But despite its ongoing popularity, there's little evidence that it has significantly changed the way that academia relates to society.

Granted, the word still has a pedagogical use. Since education remains structured in terms of disciplines – instead of in terms of problems, or judged in terms of its contribution to a vision of the good life ² – these divisions need to be reassembled. Students need to understand the relations between things, especially between science and technology and more humanistic concerns. Simpler terms would suffice, but if people want to call the making of these connections ‘interdisciplinarity’ there's no need to make a fuss.

Nor is there anything wrong with the idea of bringing in different perspectives into research. Interdisciplinarity is fine as far as it goes.

On a theoretical level, however, in terms of understanding the future of knowledge, the focus on interdisciplinarity has hampered our ability to see the challenges that lie before us. The term is a token of a historical moment in the life of the university and in university-society relations. While well-intentioned, interdisciplinarity is a half measure ill-suited to the task assigned to it. In addition, the task itself – increasing the relevance of academic knowledge – has become questionable.

The offspring of disciplinarity, interdisciplinarity has had little success in moving beyond the limitations of the former. This is largely because disciplinary knowledge isn’t structured to be relevant. In fact, for both epistemic and political reasons the opposite is the case. Concerns with interdisciplinarity will disappear as the assumptions and conditions that have supported the dominance of a disciplinary approach to knowledge fade from the scene.

To be clear, disciplines will continue to exist, as will attempts to blend their insights. Disciplines will continue if for no other reason than one cannot study everything at the same time. But the age of disciplinarity – defined by a flat academic structure, a laissez faire attitude toward knowledge production, and the separation of knowledge production from its use – will fade. Universities will take on new structures and will engage in new tasks. The shift is likely to be as significant as the one that occurred 150 years ago, which saw the transition from the medieval university and the early American college to the modern research university.

Overall, it is likely to be a diminished future. What else can one expect, since the university is no longer the premier epistemic institution of society? The professorate will likely lose much of its vaunted autonomy. Blue-sky research is likely to wither, especially within the humanities, a result that academics will despise. (This analysis should not be confused with approval.) University research will become more ends directed, as is already the case with private industry, even as private industry (e.g., Silicon Valley) will use its research and innovations to reconfigure our political institutions. One also wonders about the viability of the three (UK) or four-year campus experience, as online options increase, public support further declines, and the costs of higher education soar.

This description follows current trendlines. But these are not written in stone: the future is still there to be shaped. Unfortunately, a university culture consisting of disciplinarians is poorly suited to imagining a post-disciplinary future. The only members of university culture tasked with thinking about the university as a whole are its administrators, who were also trained as disciplinarians and lack both the time and the training for asking fundamental questions about knowledge. Similarly, two other campus units that could help plot a new course for the future of the university, philosophy departments and colleges of education, remain immersed in a disciplinary ethos.

One can imagine another future, where knowledge once again becomes hierarchical, university life is recentered around a reformed and revitalised humanities, and universities function as the governing organ for knowledge production. More on this below.

3

The central assumption of the disciplinary university is the belief that knowledge is inherently good. From this a variety of consequences follow, all of which implicate interdisciplinarity. Interdisciplinary efforts have failed because they have not questioned the assumptions that circumscribe disciplinary research.

It has been thought to be acceptable, and indeed laudable, for academics to pursue curiosity-driven or blue-sky research of whatever type they desire. This has amounted to a theological belief: as Christianity lost its hold on the educated classes in the wake of Darwin the pursuit of Truth replaced the search for God. Universities evolved out of cathedral schools, and the research university has retained some of the qualities of a cloistered space. The ivy-covered walls block unwanted intrusions and underwrite the purity of one's work. This view of knowledge has extended beyond the walls of the academy: during my time at the US Geological Survey, USGS scientists would speak of ‘throwing their science over the wall’. When I eventually asked, ‘what wall?’ I was told, ‘the one that separates science from society’.

There are real advantages to this arrangement. One convenient fact of a wall is that it blocks sight. Researchers throwing knowledge over the wall have no idea where it will land. In fact, at the USGS, directing knowledge toward a particular use or user was considered bad form. It raised the dreaded possibility of ‘advocacy’. Instead, living behind the wall meant that researchers bore no responsibility for the use their insights were put to. Those in the wider world could treat knowledge as a found object and use it as they saw fit. In this sense, all academic knowledge is technical and neutral in nature, consisting of means disconnected from ends.

But there was more: disciplinary knowledge somehow managed to be both neutral and benign in nature – neutral, in that one should not blame researchers for the social consequences of their pursuit of knowledge, while also benign, because knowledge (like God) is essentially good. The uses that insights are put to automatically count as progress. If people are hurt as a result – jobs lost, addictions incurred, communities fragmented – it is not the fault of the researcher. In any case, any injuries will eventually be recouped through the creation of still more knowledge.

I was introduced to the complexities (and contradictions) of these views some years ago while helping the National Science Foundation rework its peer review process for judging submitted proposals. Under congressional pressure, the NSF had added a second criterion to their review process, which they called ‘broader impact’. Broader impact was the complement to the first criterion of ‘intellectual merit’. Intellectual merit was seen as the disciplinary criterion – e.g., biologists evaluating biological proposals by criteria inherent to biology – while broader impact was the interdisciplinary (although more accurately in terms of aspirations, transdisciplinary) criterion for taking societal benefit into account.

At one point I suggested that since applicants were being asked to identify the potentially positive aspects of their research, they should be invited to imagine the possible negative aspects as well. I even proposed a name for potentially negative impacts – they could be known as ‘grimpacts’. The comment was greeted with stony silence. Knowledge could be beneficial or neutral, depending on rhetorical need; but it could not be acknowledged as possibly negative in its outcomes.

Interdisciplinarity has been caught in these contradictions. The raison d'être of interdisciplinarity lies in its ability to increase the relevance of academic work for societal needs. But the term is literally intra-academic in focus: all it promises is to blend different academic disciplines. This is no verbal accident. As with disciplinary knowledge, the assumption is that such blending will be ipso facto relevant to the wider world. No extra step of working with societal entities to understand their needs was necessary. In fact, when I’ve suggested to prominent interdisciplinary scholars that they work with user groups they have responded, ‘why would I want to dumb down my work for outsiders?’

The (inter)disciplinary wall has thus remained firmly in place. Its existence not only weakens the likelihood that academic knowledge will be socially relevant; it also implies that the criteria for judging academic work, including interdisciplinary work, will remain disciplinary in nature. One achieves success in academia (being hired, getting tenure and promotion, and academic renown) by living up to disciplinary standards – writing recondite articles and books that burrow ever more deeply into a discipline – even if the purported goal is societal relevance. One is thus able to talk about being relevant without engaging in the frustrating and time-consuming work of actually being relevant.

This is the space occupied by interdisciplinary work. An enterprising scholar may reach outside their discipline to bring the evidence, tools, or perspectives of another discipline to bear upon the first. The result is topical interdisciplinarity, which makes up most interdisciplinary work. Occasionally the importation becomes so successful that it gives birth to a new discipline – which then embraces the standards characteristic of other disciplines. This has been the fate of interdisciplinary studies. It has become its own discipline, with its own journals and conferences, where scholars refine their arguments without being interrupted by the demands of outsiders.

Interdisciplinarity in anything more than the topical sense has been defeated by the disciplinary assumptions of the university. Expertise is the coin of the academic realm, which perpetually places the aspiring interdisciplinary scholar in a position of dilettantism. This is a bit cockeyed, for expertise is as much a sociological as an epistemic concept. Advances in knowledge mean that standards of expertise are constantly changing, reflecting the amount of societal attention and money that has been invested in a topic. Nor is there an inherent reason why having a nuanced understanding of two or more fields shouldn’t count for as much as possessing deeper knowledge in one field.

Nonetheless, this has remained a reigning prejudice. Some years ago at a job talk I noted that knowledge has a horizontal as well as a vertical dimension and that disciplinary boundaries are largely a matter of convention. I then suggested that two masters degrees should count as the equivalent of a PhD. The comment was not well-received (and the job went to someone else). It's the vertical dimension of knowledge that counts. This despite the work of Philip Tetlock, who showed that generalists had as good or better a record at prediction as experts in a field (Tetlock 2005).

4

Interdisciplinarity has a preeminent goal – blending disciplines to improve the societal relevance of academic knowledge. Such blending often happens as a matter of course in research, since disciplines largely belong to the undergraduate textbook experience. What's more, most of the blending that occurs is small bore in nature. The real (but rare) work of interdisciplinarity consists of confrontations between the discoveries and creations of science and technology and the questions that the humanities ask concerning ethics and meaning. But more to the point, the goal of interdisciplinarity rests upon a non sequitur: combining disciplines is no assurance of relevance.

Interdisciplinary studies – i.e., the study of interdisciplinarity itself – became well organised circa 1990. Bill Newell, a central figure in the field, was trained in economics. The subject would have been better served if it had been led by people with a background in the history of ideas and conversant with knowledge production in its institutional context. In seeking to make knowledge more relevant, interdisciplinarians repeated the disciplinary assumption that knowledge is essentially benign. But just as blending is no assurance of relevance, relevance is no guarantor of goodness. Knowledge can be relevant – and even popular and profitable – while still being destructive.

Which raises the question of the relation between knowledge production and societal health. Does new knowledge always contribute to human welfare? For instance, has the research that led to the creation of the smart phone been a net positive for society? Has gain of function research on coronaviruses led to good outcomes? Have the software advances that gave us Web 2.0, and thus Facebook, Twitter, and Instagram, contributed to general well-being, or have they been simultaneously beguiling and destructive? In the case of artificial intelligence, there's concern that its creation will lead to massive social dislocation: a recent US congressional study suggested that AI could eliminate up to 100 million jobs. In October of 2025 Amazon announced plans to replace 600,000 jobs with robots (Weise 2025). But there have been no serious efforts to stop the development of AI until we get a better handle on its possible effects. ³

There's a fundamental reason that the relation of knowledge to societal health goes unasked. The assumption that knowledge is an unalloyed good, as well as our libertarian priors concerning research (cf. ‘academic freedom’) and doubts in our ability to control knowledge production are all rooted in a single assumption: questions of value aren’t amenable to rational adjudication.

Questions of common welfare are typically framed in terms of economic growth: additional wealth is one of the few values that garner general agreement. But from a historical perspective, this is to define welfare in a remarkably impoverished way. The matter of individual and societal health – aka the question of the good life – was once central to political philosophy. The abandonment of the question by the moderns was driven by a change in philosophical anthropology. Machiavelli and Hobbes viewed humans as irredeemably quarrelsome and self-interested. Contemporary history supported their view: the age of the Medicis and the shock of the Thirty Years War emphasised the worst of human nature. The combination of rising democratic sentiments and new environmental conditions – the discovery of a whole New World of resources – caused seventeenth century thinkers to discard the belief that a community should or even could work out a common set of values.

Classical liberalism (notably, John Locke) thus privatised the question of the good. Individuals should be left free to choose their own goals in life. To the degree possible, the state should be a neutral arbiter between different ends. And with values defined as subjective, knowledge production was reduced to the status of providing neutral means to private ends. Consequently, the modern university has taken a smorgasbord approach to knowledge – academics freely produce a banquet of knowledge, and people select the parts that appeal to them for whatever task they find of interest.

The longstanding belief in the irreconcilability of values has only grown stronger in recent years thanks to the rise of competitors to academic knowledge, social media and the internet. Academics disdain both as a non-peer reviewed epistemic cesspool, but this does not change the fact that the web has become our dominant epistemic space. It's been a one-sided competition: academics are simply too slow and unsavvy to compete with the flash of the internet and the virality of TikTok-generated insights or to adequately respond to takedowns that exploit the inevitable errors of expert opinion.

The web has radically democratised knowledge. On the side of content, the price of entry is minimal: rather than years spent acquiring a PhD, a couple hundred dollars provides a professional-looking website where one can opine to their heart's content. On the side of use, Google anything and results instantly appear – no need to drive to a library or read a book or make an appointment with an expert. This has given birth to a new populist epistemology where people ‘do their own research’. In this brave new world, it's unexceptional for an NFL quarterback to opine on the effectiveness of the covid vaccine. The creation of ChatGPT in 2022 provided a more authoritative gloss to what are essentially crowd-sourced opinions provided by large language models.

As a result, doubts have grown concerning the very possibility of knowledge. The resulting epistemic free-for-all has given birth to what Steve Fuller calls a post-truth society (Fuller 2018). Debates increasingly tip into epistemic nihilism, spurring political conflict, as the rational means for adjudicating disagreements has vanished. Conversations descend into verbal gymnastics offered in bad faith. The lack of consequences for insulting someone online – what was known in the early days of email as ‘flaming’ – encourages harsher public norms of behavior, which has now morphed into the automated rage of algorithms that select opinions for the controversy they generate to garner hits and attention which can be monetised. The result is a mediascape as gerrymandered as our politics.

5

We live in an era of endemic societal discord – technology-driven conflict built upon a liberal norm of value pluralism that reaches back centuries. We live with laissez faire knowledge production because of the lack of any prospect for coming to a consensus on what should and what should not be pursued. Or not quite laissez faire: knowledge production is regulated by restrictions tied to public funding mechanisms and by the choice of one topic rather than another, and also by proscriptions against certain types of research (e.g., human subjects). But the restrictions apply much less to research funded by private money, and even these are often violated with impunity given our libertarian societal norms.

This does not mean that the current state of epistemic and political conflict is a permanent condition. The internet could be tamed (see China), or conversely it might become so malign that society demands its governance, as Australia has recently put in place for those under 16. There is also the possibility that external events will emphasise commonalities within and across cultures, creating consensus around one or a few overriding values.

I see two issues that could move societies toward a new consensus on values: knowledge production and environmental calamity. Either of these – or both, combined – could inaugurate an era where common values become paramount. This would mark the end of one of the central elements of modernity. These issues could also provide a new orientation and purpose for universities. To make my point clear: rather than the continued laissez faire pursuit of knowledge and technology, I am raising the possibility that the governance, regulation, and restriction of knowledge will become a new cultural norm, which could in turn lead to the reshaping of the role of universities in culture.

This brings my argument in alignment with contemporary events. The striking political shift in Silicon Valley in the 2024 election, where many leading figures turned from the Democrats to Trump, was in large part a response to Democratic attempts to regulate technology. Parts of Silicon Valley put this shift in apocalyptic terms: Peter Thiel has described those who want to regulate technology as Luddites in league with (or perhaps are) the Antichrist. But even calmer voices within that community abhor the idea that technology should be regulated. This highlights a fundamental disagreement about the future of our epistemic efforts. Some fear that unbridled knowledge production will bring on the apocalypse, while others claim that any restraints upon knowledge and technology will be the cause of that apocalypse.

I will leave it to others to advance the latter thesis. Under the former possibility, consider first the case of unbridled knowledge production as the engine of additional technological advance. There are limits to technology – if from nothing else, from physical laws like the speed of light and the intrinsic capacities of atoms and the materials that they form. But from the perspective of human capacities, technological development is infinite. This disjunct – between the number and speed of things and the human pace of life – can be overwhelming.

The pace of human activities varies, of course. We enjoy both fast food and a leisurely dinner. And our busy schedule may cause us to try for quality time with a child or sick relative. But there are temporal limits to such activities if they are to remain human-scaled. Few of us would be satisfied if eating consisted of popping a pill, nor are we able to eat for days on end. Similarly, there are limits to how short a time spent with a sick relative counts as doing anything for them at all.

We increasingly inhabit environments whose speeds, number of options, and degree of attractiveness are out of sync with our species being. We are caught within the mismatch between human pacing and the velocity of electrons, struggling to triage the daily onslaught of information, choices, and cultural churn. And the gap between technology and human capacity continues to grow.

What links the prevalence of obesity, our attachment to smart phones, the incel's immersion in internet culture, people's inability to control the use of credit cards, the opioid crisis, and the rise of right-wing populist parties? All are reflections of the mismatch between the temptations created by technology and our ability to exercise self-control. Of course, not everyone is fat – and now there is a drug for that – and many people resist the temptation of drugs, the blandishments of advertising, and attempts at spreading misinformation. But the present attractiveness of these technologies is temporary. Driven by the profit motive, companies are in a constant race to increase the seductiveness of their products.

All of which serves as a spur to populist politics. Accelerating technological development aggravates the problem of cultural lag. People failing to keep up with the rate of technological change often find comfort in the explanations of demagogues who blame the dislocations in their lives on foreigners, minorities, or corrupt elites. Technology-driven disruption thus provides opportunities that authoritarians turn to their own ends.

Evermore powerful types of addiction is not the only problem raised by technological advance. Whether through the spread of CCTVs, the increased use of biometrics, or the voluntary (and often unknowing) sharing of private information, governments and corporations are in the position to know the whereabouts and the activities of all its citizens. Citizens are reduced to hoping that both will restrain themselves from using the tools that they now possess for controlling the population.

Second, consider the potential crises tied to the environment. There is already (albeit presently a largely ineffectual) global consensus on the seriousness of climate change. Droughts, heat waves, and other types of environmental disruption may heighten the salience of other environmental issues such as the loss of pollinators, the presence of forever plastics, or the acidification of the oceans. Of course, environmentalists have been predicting ecological collapse for years. But their concerns may not have been so much wrong as premature. A major calamity could prompt the forming of a societal consensus around the necessity of considering the environmental consequences of new epistemic projects. ‘Health’ could then become the banner around which a consensus forms, as society expands its definition to include mental health (to better manage the social effects of new technologies) and the health of the environment (to e.g., safeguard basics like water and food production).

6

In an era of constant change, we still forget that things change. It's possible for a culture, even one as divisive as ours, to move from 300 years of value pluralism to a consensus on values. And so it's worth considering how universities could respond to such circumstances.

The formation of societal norms concerning core values could affect universities in two ways: by structuring teaching and research with an eye toward advancing these societal goals, and by helping to identify appropriate limits to knowledge production. In the former case, we can learn from historical precedent – the medieval university. The modern university is flat in the sense that no discipline is subordinate to any other and there is no general end that knowledge production serves. Both knowledge production and use are essentially libertarian. In contrast, medieval universities were hierarchical in nature. While having a reputation for being intellectually stultifying, medieval universities advanced knowledge in various ways – through the study and translation of classical Greek and Arabic works, by placing emphasis on logic, mathematics, and natural philosophy, and by serving as a crucial incubator for the scientific thought. At the same time, epistemic innovation remained subordinate to the task of passing on a fragile cultural legacy to the next generation.

That cultural legacy was religious in nature. Knowledge was hierarchically arranged in order to promote the societal goals of saving one's soul and to promote a Christian community. This did not preclude other goals: after all, it was the need for people to keep the church's books that led to the development of universities out of the cathedral schools. But these subsidiary goals were not to come in conflict with the needs of the Christian community.

Medieval universities were organised in terms of lower and higher faculties. The lower (humanities) faculty were allowed a freer mandate while being preparatory for the higher faculties of medicine, law, and theology. The three higher faculties formed their own hierarchy, structured in terms of an ascent from medicine (care of one's body) to law (care for the body politic) to theology (care for one's and the community's soul).

To state the obvious, I am not advocating the reestablishment of a Christian polity. Rather, the point is to reconnect knowledge production to a common vision of the good life. Agreement on such a goal would prompt knowledge to be restructured as a hierarchy, with questions of ethics and value – that is, the humanities – at the apex. Researchers across the disciplines would still be able to pursue any number of projects, but these research paths would be subject to a paramount consideration: what implications does a given line of research have for the overall goal of societal and environmental health? ⁴ In many cases these criteria would have no effect on research. But there would be cases where these criteria would modify, redirect, or even stop the initiation of research.

Which highlights the second role of the post-disciplinary university: assisting in the governance of knowledge. As the home of epistemic expertise, and in partnership with government, universities would be the natural site of knowledge governance. Universities have a nascent version of such governance today in the existence of institutional review boards. IRBs review prospective research to ensure that it complies with applicable regulations and ethical standards, especially regarding human subject research. These boards now focus on only a portion of research internal to the university, but they could be expanded to review all research.

This would address government-funded research, but what of private research? Much of the most problematic research today is conducted by corporations. Universities could also be the site for the review of this research, but this would require laws that obliged corporations to describe their current and anticipated research with the reconstituted and expanded IRBs. A nascent form of such review already exists in the reliance of governments upon the work of committees consisting of academic experts for, e.g., the drawing up of legislation. But make no mistake: the governance of private research raises a range of political, economic, and practical challenges.

Exploring the possibility of a system of governance is the work of another essay. But to note the obvious, there will be individuals, corporations, and nations that will try to evade any review of their activities. All laws are imperfectly enforced, but attempts to govern technological advance will face distinctive problems. Knowledge has been radically democratised by the internet; how could one possibly control all its dark corners? And how could we control nations who will be tempted to seek political and military advantage via innovation, a scenario that we already see surrounding the development of AI?

These are vexing problems. My melancholy suspicion is that it will take a catastrophe resulting from technological advance for nations to summon the will to regulate knowledge. Even then, exercising any real control over the production of knowledge would require the instantiation of an entirely new cultural project: rather than the modernist goal of continually augmenting our powers via science and technology, something like a post-modern project in Buddhism, fostering a culture of maturity and self-control. If one dared to hope, figures like Trump and Musk would come to be viewed as exemplars of egotism and greed that have been consigned to history.