Future Diplomat

Reimagining Diplomacy for 2035.

Concept rendering of a British “self-driving embassy,” an AI-operated mission built around data-center infrastructure.

Future Diplomat: Rewiring & Reimagining Diplomacy

Technology has long changed diplomacy. If diplomacy is, according to the Oxford English Dictionary:

- the management of international relations by negotiation; the method by which these relations are managed by ambassadors etc; skill in such affairs. Adroitness in personal relations; tact.

… then diplomacy is communication: both the channel by which states speak, and the craft by which they persuade. As technology revolutionises how we communicate, to make and shape decisions, so it revolutionises the conduct of diplomacy.

Think of diplomats, embassies and the FCDO as cybernetic systems. They seek and sense information, process it through the neurons of the brain as individuals, or between the functional roles of the people staffing the organisation, act or choose not to, then observe the effects. The end is behavioural: get someone to do something new, do something differently, or keep doing what they already do. It might be a foreign person, organisation or group is the target of that diplomacy, or it might be those within the diplomat’s organisation, their home country or government – where the diplomat is seeking to mediate between positions.

If this is true, it should be clear that technologies that are fundamentally informational, and cognitive – they draw insight and foresight from data at unprecedented speed and scale, and make decisions based on models of the future – will transform diplomacy.

This vision for 2035 is built on the premises above. In addition, its starting assumption is that using forecasting tournaments and/or prediction markets¹ is the best way to understand the future – on which basis it anticipates:

- weak artificial general intelligence (AGI), one that matches or exceeds human performance on a range of diverse cognitive and informational benchmarks, by March 2027²;

- Oracle AGI, a system that significantly exceeds human performance in answering any question, by 2028³ – or under a more stringent definition 2032.⁴

- Artificial Superintelligence (ASI) – adding superhuman skill in robotics, proprioception, dexterity, locomotion ~May 2029⁵;

Thus a diplomat in 2035 will be operating in world of superintelligent AI.

Consular Services Without Counters

Most consular work is structured information processing: visas, passports, notarial services, births and deaths, crisis assistance. By the early 2030s, agentic systems will handle the bulk of it end‑to‑end with clear audit trails and explainable reasoning. Consulates will be able to operate in all languages, tailoring information to individual cultural and psychometric dispositions and levels of education. Identity will be verified remotely through secure biometrics. Citizens will complete interviews and attestations conversationally, recorded in signed digital form. There will be little need to visit consulates for these tasks, with most activity taking place online, and other interactions via the metaverse – meetings with AI (probably ASI, per current forecasts) agents embodied in avatars. If these are distinguishable from a remote meeting with a human today it would only be in that we should expect an advanced, or superintelligent AI to provide a better service than a human – more empathetic, more patience, more personalised, smarter, more knowledgeable, with better memory.

From Self‑Driving Labs to Self‑Driving Embassies

In 2023 Eric Schmidt, in MIT Tech Review, & the scientific journal ‘Nature’ both described the birth of self-driving labs (SDL), those that - at the maximum extent - operate fully autonomously, designing hypotheses and experiments, conducting experiments, reporting the results, and then designing new hypotheses, testing again, without human intervention, at super-human speeds. In 2025, self-driving labs are driving breakthroughs globally. The journal Nature Chemical Engineering reports that in the field of materials science self-driving labs are producing new materials, ten times faster, processing ten times the data, accelerating discovery exponentially. Furthermore, July 2025 saw Nature publish articles showcasing:

1. a ‘self-driving microscope’ to better understand protein aggregation and neurodegenerative disease, and

2. “Ada” a self-driving lab that has discover new metal materials by identifying optimal trade-offs between conflicting objectives

3. Showcasing a Chinese lab’s use of AI to design the ‘brain’ of a quantum computer using laser light to move thousands of atoms into any arrangement they choose and act as qubit, in 60 milliseconds - demo’d in this case to produce a video of Schrodinger’s cat.

While the same month this year (2025) Science reported how self-supervised AI now gives robotic probes pixel-level precision, letting autonomous labs log over 3,000 semiconductor tests in a single day - enabling rapid defect detection and a stronger, more resilient advanced-manufacturing supply chain. Given AI Agents are reportedly doubling their performance ever seven months, these kind of breakthroughs are likely just the beginning (see Cassi blog post dated 31 July for more SDL progress in 2025).

The next decade, perhaps the next few years, will see the emergence of ‘self-driving embassies’. Just as today’s self-driving labs use autonomous AI to answer scientific questions, in self-driving embassies AI agents will handle consular and diplomatic enquiries and AI systems mediate most—perhaps all—routine exchanges, principally text, audio or digital, between embassy and both home and host governments.

Persuasion agents will build psychometric profiles of the individuals and groups with whom the Embassy or diplomat is interacting, enabling tailored negotiation strategies, which can be adapted and improved in real-time. When negotiations are in-person via video call, AR/VR meet, or physically face-to-face, ASI agents will be reading micro-expressions, designing recursive negotiation strategies and updating the approach continuously. Affective computing will be more empathetic than humans, able to understand and manipulate us at superhuman levels. It is hard, though not impossible, to imagine negotiation being fully delegated to ASI systems negotiating with each other. A study in Nature in May 2025 showed AI was more persuasive in debate than humans >60% of the time. A recent (2025) unauthorised and probably unethical experiment on the Reddit forum R/Change My Mind shows the way – an LLM deployed to persuade was six times more persuasive than humans and operated without anyone recognising it was an LLM. The largest LLM persuasion study to date, (published 18 July 2025) showed newer, larger models were the most persuasive, and that models persuasiveness persisted over time (1-month, in this study). The best models (already a generation behind o3 and GPT-5) were 41% and 52% more persuasive than ‘static’ messaging - remember when we were panicking about ‘static’ social media messages on Twitter and Facebook? In fact, previous studies of human persuasiveness, advertising and personal conversations in elections, suggest ‘an average effect of zero’.

Even if human relationships remain crucial to effective influence and negotiation, they will be increasingly informed by profiling agents, prompting and preparing diplomats for discussions, perhaps advising during discussion, debriefing afterwards.

Linguistic skill may be vital, not for translation - which may cease to be relevant when translation can be automated instantly - but to enable diplomats to develop the mental models and worldview of those they interact with. If that is the case, diplomats will likely be learning through AI-tailored immersive courses in AR and VR. Cultural expertise, and understanding - the right-brained process still too little understood by neuroscientists, psychologists, computer and other scientists for us to be confident we are modelling it effectively in Artificial Intelligence, may be a key human skill. But to outperform or complement machines it will likely need longer, deeper immersion in country, possibly career-long deployments. Similarly, when the human currency becomes relationships, since rationality, persuasion, information processing and reporting is undertaken at superhuman level by machines, FCDO careers will be much more focused on the cultivation of life-long relationships with current and future key influencers in countries around the world. Fast-stream processes, and civil service churn will have to end if diplomats are to flourish in an age of AI. Deep relationships take time to build.

Recruitment of diplomats will use AI systems to identify super-persuaders, just as intelligence analysts will be identified as super-forecasters, or super-questioners – those that can create the questions that maximally reduce ignorance, on which forecasters and ‘persuaders’ should focus – and recruitment will follow. A more prosaic instantiation of Philip K Dick’s Precogs in ‘Minority Report’. Strategy, tactics, decisions will be automated. Humans will be working out what the most desirable ‘outcome’ or objective is with machines, but the planning itself will be done by machines, and implementation overseen and directed by them.

Passports to New Worlds

The trend towards e-citizenship and e-businesses will continue. Many visa and passport applications will be for virtual access to the UK, as telepresent (AR/VR) or telexistent agents (AR/VR with the ability to directly interface with the real world via robotic interfaces) – with people ‘visiting’ and working in the UK without ever entering the country.

Furthermore, citizens and businesses will apply for Visa to access UK managed virtual worlds, metaverse countries with economies that rival countries in GDP with access to UK legal authority for dispute resolution.

Beyond the ‘Geo’ in Geo-Politics

Migration pressures will ease, as citizens in developing countries are able to prosper by building e-businesses in developed countries while remaining in their own. Trade barriers and cross-border taxation will become harder to enforce as metaverse economies produce wealth independent of geography.

Space will be the most lucrative market in human history, with in-space industries from mining to manufacture, compute to colonisation. By 2035, with regular vertical and horizontal launch from the UK, the FCDO will have to manage in-space territorial claims and economic competition, as well as the security implications that flow from competition for dominance in space. If the UK fails to compete in the space colonisation, asteroid mining and space-based industries, the FCDO will be managing a much more precipitate decline in the UK’s relative power in the international system than it has ever had to do to date.

Finally, automation, robotics and AI will reduce countries and companies' dependence on demography to mine raw materials, produce goods, services, and scale armed forces to deter or fight. Self-driving labs will become self-driving innovation systems, with companies and countries power, prosperity and security driven by the speed at which they can invent and harness new technologies.

In combination, metaverse economies, the space market and automated innovation and industries will move competition for relative power in the international system beyond the geo of geopolitics, with nations’ within-border natural resources and demographics less deterministic of their power and influence. For the FCDO, effective statecraft will have to be more integrated x-Whitehall, and with industry, finance and society, than ever before.

Influence, Security, Privacy, Agency

Superintelligent systems able to analyse groups and individuals, understand their psychometric make-up, and influence their behaviour risks social control tightly circumscribing human agency, able to manipulate with unprecedented effectiveness. Even today, we generate c. 1.7MB of data per person per second – the equivalent of 56 million Kindle books per on each of us generated every day – it is this that gives unprecedented insight and foresight into who we are, what we do, and what we are likely to do in the future. Debates on human agency will soon be the domain of the diplomat, as privacy-centric EU clashes with the market-centric US, and China’s digital authoritarianism.

Our security will be redefined in a world where the boundaries of the virtual and the physical have blurred through AR/VR, where we live in different ‘belief circles’ seeing different versions of the world around us to our neighbours, and spending more and more time in borderless cyberspace. Cognitive security may be a competition for e-citizenship, e-businesses with other states, and virtual states.

When our cars, smart homes and brain computer interfaces all become critical national infrastructure through connected IoT devices, we will have to rethink how it is managed. Cognitive security defending our data and our minds.

All this will make the role of the FCDO more complex, and more expansive, while the future diplomat might be able to more effortlessly cover a wider geographic area, the GeoInt Singularity + ubiquitous data surveillance will mean they will struggle to move anywhere or meet anyone, without others knowing. It may be that everyone they meet knows their psychometric profile, their preferences, and has optimised their position to ‘win’ in diplomatic negotiation.

Perhaps privacy-enhancing technologies will prevent this, but if they do, they will also limit the state’s ability to manage threats, gather data.

Public diplomacy conducted online will become an ever more powerful tool, but will also raise the same questions of privacy, security and agency – how far should we go in profiling and manipulating in a world where Spotify data can be a better indicator of national mood than direct survey data⁶, where all of us are vulnerable to microtargeted manipulation.

In this context, regulatory diplomacy will be as important as all other forms, determining the kind of world we live in, and the extent to which people are free. Hence China requires all AI companies to design with “core socialist values”, values it seeks to embed and export via its proposed ‘Organisation for Global AI Cooperation’ (Reuters, July 2025; Chinese State Council, 26 July).

Simulation and Wargaming

New policies, new negotiating tactics, the effect of military or economic actions, likely future scenarios and emerging flashpoints we all be constantly developed and tested through the digital strategy-engines, offering superhuman suggestions for strategies and tactics, testing second order effects, highlighting new factors to be considered. The GeoInt Singularity, where every square meter of the earth is constantly observed from space, will enable the development of a global digital twin, to identify flashpoints and early indicators of emergent conflict or natural disaster. Wargames and collective intelligence tools will enable humans to challenge the models assumptions, force it to consider counter-factuals, ensuring the most effective strategies are pursued.

Augmentation

Technological transformation will reduce the role of humans in many diplomatic decisions, and reduce the number needed to deliver diplomatic services. However, human relationships may remain key to successful negotiation - it depends on the extent to which nations are willing to let AI-agent negotiate with AI agent. If humans remain in the loop they will be increasingly augmented, able to bring the best of human cognition - whatever remains of value in this emerging age – together with technology to enhance our capacities. Brain-computer interfaces will, by 2035, likely be enhancing memory, perception and learning. In turn we will be aided by a neuropharmacology industry that optimises our performance and our rest, and by a growing science of high-performance teams, that ensure we best balance cohesion, talents, and incentives to maximise the performance of people in groups.

The US AI Umbrella?

If the US develops an AGI or ASI model before anyone else, is able to harness it effectively, and launch it first on recursive self-improvement, it may be that no other nation can catch-up. In such a situation, British and other nations diplomats would be negotiating with a system that was comprehensively smarter than them, and progressively more so – those they interact with being prepared not by very capable advisors, but by a superintelligence Private Office or equivalent – or possibly being asked to interact directly with the US’ ASI diplomat. So too British businesses would be facing competition from businesses smarter than them, the British military working with an ally, assuming the US remains such, that was strategically, operationally and tactically smarter. The outcome of such competition is likely to be very one-sided. If this becomes so, we may find ourselves living under the AI equivalent of the US nuclear umbrella, surrendering sovereignty, agency, in return for protection and access. The price of living under the US AI Umbrella is likely to be far higher than the cost of its nuclear protection.

AI Disruption or Destruction

It could be that we live, as Oxford Philosopher Nick Bostrom puts it in a ‘solved world’ a post-ASI utopia where nobody need concern themselves with the intra-human competition of which diplomacy is part. It might be that through misalignment or malignance AI poses an existential threat to humanity. Or it might be that AI becomes a substitute, not a complement to human labour, with radically disruptive social and political effects. None of these scenarios are so unlikely that the can be ignored. Hope is not a strategy. But this article has described the Future Diplomat operating in a messy, complex, competitive world – with or without the development of super-intelligent AI. It may be unproductive and distracting to spend too much time on the dystopian outcomes here. It is vital the FCDO does spend time on them, nonetheless.

Conclusion

Much of what has been described is possible now, with today’s relatively narrow AI. More will be possible as current AI advances further - all the evidence suggests the UK will underestimate this progress, and the nation be poorer and less secure as a result.

In the more likely than not advent of AGI or ASI before 2035 the role of the diplomat and Embassy will be radically transformed, with humans the agents of AI as much as the other way around. The birth of metaverse worlds and the growing import of the space economy by 2035 will change where diplomacy is conducted, power created and managed, people and machines persuaded. We must be ready.

The best time to start building for this future was yesterday. The next best time is today.

1. For a good overview of the difference, see: https://www.metaculus.com/notebooks/38198/metaculus-and-markets-whats-the-difference/

2. Metaculus, 2025. Date Weakly General AI is Publicly Known. Accessed 31 July 2025.

3. Metaculus, 2025. Oracle AGI Precedes General ASI. (20 months). Accessed 31 July 2025;

4. Metaculus, 2025. Date of Artificial General Intelligence. Accessed 31 July 2025.

5. Metaculus, 2025. Time From (weak) AGI to Superintelligence (25.9 months). Accessed 31 July 2025.

6. See e.g.

a. Edmans, Alex and Fernandez-Perez, Adrian and Garel, Alexandre and Indriawan, Ivan, Music Sentiment and Stock Returns Around the World (August 14, 2021). Journal of Financial Economics (JFE). Available at SSRN:https://ssrn.com/abstract=3776071 or http://dx.doi.org/10.2139/ssrn.3776071.

b. Kaivanto, K. and Zhang, P., 2019. Popular music, sentiment, and noise trading. Lancaster University Management School.

c. Sabouni, Hisam. (2018). The Rhythm of Markets. 10.13140/RG.2.2.31484.64646. Pre-Print https://www.researchgate.net/publication/323560860_The_Rhythm_of_Markets