Five Dispositions for Future Public Officers

The veteran public sector practitioner, observer and CSC Fellow argues that an increasingly turbulent world calls for the intentional formation of human habits – of mind and action – that cannot be automated away. 

Home as a Place We have Never Been 

An anniversary is an invitation to re-read oneself – a less comfortable exercise than it sounds. Preparing this essay, I reviewed the pieces I have written for ETHOS since 2011, expecting to find positions I had outgrown. Mostly, I found the opposite: arguments that still seem broadly right, made with a confidence I would now temper, about a world that has arrived faster and stranger than I anticipated. Le Guin's paradox describes the sensation exactly. The words on the page have not moved. The place they describe has. 

What has shifted for me is less the content than the underpinning concepts. In 2011, writing about preparing public officers for new modes of governance, I thought I was writing about skills and capabilities that could be specified, taught, and assessed. I now realise that I was writing, without naming it, about formation: the slower work of shaping officers as whole persons: what they perceive, attend to, are willing to sit with, who they become over the course of a career. Skills are acquired. Formation happens, well or badly, whether or not an institution intends it. 

This matters now because of the particular pressures on public services. Artificial intelligence and digitalisation are the most conspicuous forces, and rightly command attention and budgets. But they are best understood against a broader context: ageing populations and shrinking workforces that make productivity acute public concerns; climate transitions that will force severe distributional decisions; geopolitical turbulence that has turned supply chains, semiconductors and sea lanes into instruments of statecraft; a fraying of shared identity in many societies that makes collective life much harder than it was a decade ago. Each of these compresses the time available for decisions while expanding the number and weight of considerations that bear on them. 

The temptation, faced with such a list, is to respond with a curriculum: modules on AI literacy, modules on climate, modules on geopolitics. In the spirit of formation rather than simply skills-building, I suggest something more difficult. Technical upgrading is necessary yet insufficient. What the next generation of public officers will need is a set of dispositions that make technical capability usable under conditions of turbulence, moral difficulty and public scrutiny. Five qualities seem particularly indispensable. 

1: Complexity-readiness 

In 2011¹, I argued that the Singapore Public Service had passed through successive developmental stages, from providing basic services and security to building adaptive and then relational institutions, and that the later stages demanded three unfamiliar capabilities: obliquity, or addressing root causes indirectly rather than attacking problems head-on; bio-empathy, or reading systems as ecologies with feedback loops rather than as machines with levers; and creative making, the instinct of homo farens, the doer, alongside the analyst.

I still believe this. But I underestimated how much of this is temperamental rather than technical. Obliquity is not really a method; it is a tolerance for delayed gratification and for the professional risk of appearing to do nothing while working on long-term causes. Bio-empathy is not a modelling technique; it is a willingness to be surprised by one's own system. Both are undermined by the very things public administrations reward: speed, decisiveness and legible attribution of results. This is why rigidly performance-led approaches to learning and development have to be handled with care. Measuring what learning produces is right and overdue. But a system that measures only what resolves within an appraisal cycle will quietly select against exactly the officers who can wrestle with a wicked problem long enough to understand it. 

Complexity-readiness is partly a curricular question, answered through simulations, policy gaming and cross-agency sense-making of the kind the Civil Service College (CSC) pioneered, and partly a question of institutional conditions: whether an officer who says "we do not yet know" is treated as rigorous or as evasive. 

2: Digital and Data Literacy 

Writing in 2019² about how governments might measure digital success, I was chiefly worried about vagueness. Indices told us whether governments had technology, not whether the technology worked; “modernity” stood in for effectiveness, and presence for usefulness. I proposed what I hoped were more honest measures, covering project completion, price competitiveness, system effectiveness, usage and usability, data security, timeliness, and whether data actually informed policy. 

Two things have deepened since. The first is a point I made in a later ETHOS conversation³ on data, policy and outcomes: the right operating model is that some people must do this work all the time, and all people must do it some of the time. Specialist data and AI teams cannot be the whole answer, because the demand side matters as much as the supply side. What determines whether evidence changes decisions is trust between parties, skills at every level, someone able to bring people around a table, genuine hunger for the evidence, and a peer group that sustains the habit. Most of these qualities cannot be procured; they must instead be nurtured. 

Second, generative AI has changed what literacy means. It is no longer mainly about interpreting output, but judging what to delegate and what to withhold; recognising the specific texture of a plausible falsehood; knowing which decisions require a human author because someone must be answerable for them. Public officers work under a standard the private sector does not face: a citizen who is refused a benefit is owed reasons, and ‘the model determined this’ is not a good enough reason. Literacy in the public sector therefore has an irreducibly human dimension, linked to an individual’s constitutional rights and the state’s concomitant obligations. 

3: New Metaphors 

Writing for this journal's issue on the future of learning⁴, I argued that our dominant metaphor for learning, the learner as receptacle to be filled, carried assumptions that no longer serve us, among them instrumentality, linearity, fixed capacity and passive receptivity. I proposed the metaphor of a garden instead: uncertain, non-linear, requiring pruning as well as growth, with organisms that shape their own conditions. 

I have since extended the argument⁵ from learning to governance itself, suggesting that networks and platforms, for all their advantages over control towers and vending machines, remain too static, too government-centric, and too transactional to carry the moral weight of public life. Governance is better imagined as a moral ecology: living, untidy, subject to cycles of decay and regeneration, and bound together by obligations rather than transactions. 

The metaphors currently on offer for AI in government are doing enormous unexamined work. “Copilot” implies a cockpit with clear instruments and a single destination. “Agent” implies delegation to something with interests of its own. “Augmentation” implies that the human task remains fixed while capacity around it expands. Like all metaphors, each highlights something real and hides something important. Officers who can notice the metaphor they have absorbed, and consciously choose another when needed, will make better decisions about where these tools belong, and be better able to communicate these decisions to an increasingly discerning public. This is among the most practical skills on my list, even if it looks the least practical. 

4: New Abundances and New Scarcities 

A decade ago I wrote about generative resources⁶: data, knowledge and relationships, which expand rather than deplete through use, suffer from underuse rather than overuse, have fuzzy boundaries, and settle at no clear equilibrium. I argued that governing them well required different instruments from those we use for oil, land and time. 

AI is among the most generative resources we have yet encountered, and it makes the argument both more urgent and more complicated. Cognitive tasks that were once scarce — drafting, summarising, translating, first-pass analysis — are becoming abundant. But abundance in one layer creates scarcity in adjacent ones, and the new scarcities are the interesting part. Verified information becomes scarce as unverified information becomes free. Attention becomes scarce as content becomes limitless. Institutional trust becomes scarce precisely when synthetic media makes it easiest to counterfeit. And in a striking return of old-school physical constraint, energy, water and land become scarce again as data centres compete with households and industry for all three. This is where the AI agenda and the climate agenda stop being separate conversations. 

The core disposition needed here is diagnostic: we need to ask, of any new capability, what it makes plentiful and what it thereby makes precious. As my earlier piece notes,  we are not facing a utopia of abundance. Scarcity and inequality persist, and generative resources tend to accrue to those already positioned to use them. 

5: Harnessing Deliberation and Engagement 

Most recently, I argued⁷ that participatory governance succeeds or fails less on institutional design than on micro-motivations, and that projects fall into four archetypes depending on whether governments and stakeholders are each willing or unwilling. I cautioned about the trap of a suboptimal Nash equilibrium in which both sides find it rational to assume the other is unwilling, and both are proved right. 

Two developments have sharpened this. Polarisation and identity-based fragmentation raise both the cost and the value of deliberation: it is harder to convene, but even more consequential when it works. AI now sits on both sides of the ledger. It can enable deliberation at genuine scale, through real-time translation, synthesis of thousands of submissions, and surfacing of the points on which people unexpectedly agree. It can also flood the same channels with synthetic participation, making it rational for officials to distrust the very inputs they solicited. 

Which way this goes will be settled by something Conway's Law anticipated: technology tends to reproduce the communication structures of the organisations that adopt it. Hierarchical systems will use AI to consult more efficiently while listening no better. That makes willingness itself, the readiness to be changed by what one hears, a capability worth developing deliberately, and one that no procurement exercise will supply. 

Connecting the Dots 

Nobel Laureate Wislawa Szymborska has a poem, “Nothing Twice”, that all public officers should probably be pointed to on their first day: 

This is the condition the five capabilities can address. Consequential decisions are taken by people who did not get a rehearsal, in circumstances that will not recur in the same configuration. No syllabus removes that. What a syllabus can do, at its best, is form people who are steadier in their decision-making. 

The five dispositions are not modules to be delivered sequentially. They are facets of a single formation, and formation requires conditions as much as content: time that is genuinely protected rather than nominally allocated, seniors who model uncertainty rather than perform omniscience, and career structures, of the sort that CSC’s Career Fitness agenda is reaching towards, that treat a public service career as something cultivated over decades rather than optimised each year. 

It is easy to think that the case for AI in the public sector is about productivity, and that the human questions are downstream constraints to be managed. The reality is the reverse. The reason to automate routines is to return officers to the work only humans can do: sitting with a distressed citizen, holding a trade-off that cannot be dissolved, taking responsibility for a judgement that will be wrong for someone. Technology's contribution to human flourishing in the public sector will be measured by how much truly human work it allows us to do, and how much it quietly takes away. 

Which brings me to the hard truth underpinning all five skills. Public officers of the future will work under conditions that no capability framework can make comfortable: climate losses that have already begun, conflicts that will not resolve within the span of a career, neighbours who no longer share a common account of reality. Formation is not preparation for a world that will eventually settle down. It is preparation for one that will never settle, where steady states are theoretical concepts at best. 

Novelist Italo Calvino puts it better than any competency framework could: 

Twenty-five years of building a public service's capacity to learn is, at its best, an exercise in Calvino’s second approach. Seeking out what is not inferno: what is precious. Making it endure, giving it space, and ultimately forming ourselves to return to homes where we have never been. 

ABOUT THE AUTHOR

Aaron Maniam is Professor in Practice at the Lee Kuan Yew School of Public Policy and Deputy Director of the Institute of Policy Studies, both at the National University of Singapore. From 2023-26, he was Fellow of Practice and Director of Digital Transformation Education at the University of Oxford's Blavatnik School of Government. He previously served in the Singapore Government, including as founding Head of the Centre for Strategic Futures and as Institute Director of the Institute for Public Service Leadership at the Civil Service College. He has been a Fellow of the Civil Service College since 2024. 

NOTES

1. Maniam, Aaron. "Preparing Public Officers for New Modes of Governance." ETHOS Issue 10 (2011). https://knowledge.csc.gov.sg/ethos-issue-10/opinion-preparing-public-officers-for-new-modes-of-governance/ (opens in new tab)

2. Maniam, Aaron. "What Digital Success Looks Like: Measuring and Evaluating Government Digitalisation." ETHOS Issue 21 (2019). https://knowledge.csc.gov.sg/ethos-issue-21/what-digital-success-looks-like-measuring-evaluating-government-digitalisation/ (opens in new tab)

3. Online conversation in Ethos Digital Issue 12, transcript available at: https://knowledge.csc.gov.sg/digital-issue-12/data-x-policy-x-outcomes/ (opens in new tab)

4. Maniam, Aaron. "Metaphors & Mental Models: Updating Our Language for Public Sector Learning." ETHOS Issue 23 (2021). https://knowledge.csc.gov.sg/ethos-issue-23/metaphors-mental-models-updating-our-language-for-public-sector-learning/ (opens in new tab)

5. Maniam, Aaron. "Beyond Control Towers, Vending Machines and Platforms: Towards More Dynamic, Living Metaphors for Governance." Policy Quarterly 20: 1 (2024). https://doi.org/10.26686/pq.v20i1.9050 (opens in new tab)

6. Maniam, Aaron. "From Scarcity to Generativity: New Approaches for Governing Resources." ETHOS Issue 16 (2016). https://knowledge.csc.gov.sg/ethos-issue-16/from-scarcity-to-generativity-new-approaches-to-governing-resources/ (opens in new tab)

7. Maniam, Aaron. "Varieties of Engagement in Government-Citizen Interactions." ETHOS Issue 28 (2025). https://knowledge.csc.gov.sg/ethos-issue-28/varieties-of-engagement-in-government-citizen-interactions/ (opens in new tab)

Back to ETHOS Issue page