■ MODERATE RISK ■ Public Service & Government
No. States are not going to outsource sovereignty to a language model, and trust between governments is still built by humans at dinner tables. What AI will replace is a large share of the cable-writing and briefing-book drudgery underneath the diplomacy.
“AI translates and analyzes, but nobody sends a chatbot to negotiate a peace treaty.”
Our AI replacement risk score — how we score jobs
Diplomacy runs on an unglamorous paper mill. Foreign service officers spend their days drafting reporting cables, preparing briefing memos for visits, managing consular caseloads, tracking host-country politics, and negotiating text — often literally arguing over a single word in a communiqué for hours. Above that sits the real product: relationships with host-government officials, journalists, and opposition figures cultivated over years, and the judgment to know when a counterpart's 'we will consider it' means yes, no, or send money.
The paper mill is exactly what large language models are good at. Summarizing local press, drafting first-pass cables and talking points, machine-translating documents and meetings, scanning treaty text for inconsistencies, war-gaming negotiation scenarios — foreign ministries are already piloting all of this, and consular sections use automated systems for visa triage at scale. A junior officer's traditional apprenticeship of synthesizing and summarizing is the layer under the most pressure, which raises a real question about how the profession trains its next generation.
The core resists for structural reasons, not sentimental ones. Negotiation between states is a trust game conducted amid deliberate deception, where a diplomat's read of a counterpart — and the personal credibility to make a quiet assurance stick — is the actual currency. No government will let an algorithm make commitments that bind the state, and much of the job's value happens in corridors, at receptions, and in one-on-ones that never appear in any dataset. Our risk score is low: the cable-writing shrinks, the statecraft doesn't.
Automatability: our editorial assessment of current and near-term AI capability
Foreign ministries are cautious adopters, but by the early 2030s AI drafting, translation, and open-source analysis will be standard tradecraft, and consular processing will be heavily automated. That thins the junior-officer paper apprenticeship and changes how diplomats are trained. The negotiating, relationship-building core remains human through ~2040 and beyond — states simply won't delegate it.
It can simulate them, red-team them, and analyze positions — and negotiating teams increasingly use it for exactly that. But actual state-to-state negotiation requires making credible commitments on behalf of a government, reading counterparts who are professionally deceptive, and building trust across years. No state will hand an algorithm the authority to bind it, which keeps humans at the table.
Yes, though the shape is changing. The traditional junior-officer workload of summarizing, drafting, and reporting is the most automatable layer, so early careers will involve less paper and more expectation of judgment, languages, and networks from day one. The senior craft — negotiation, representation, crisis management — looks durable for decades.
Mostly in the back office: machine translation, summarizing foreign media, drafting first-pass cables and talking points, scanning treaty language, and triaging visa applications at volume. Several ministries also use AI for open-source intelligence and negotiation preparation. The pattern is consistent — AI feeds the diplomat, it doesn't sit in the meeting.
Fluency with AI analysis tools, because officers who can direct and verify machine-generated synthesis will outpace those who can't. Beyond that, the classic assets appreciate: hard languages, deep regional networks, and increasingly technology policy itself, since AI governance, export controls, and digital trade are becoming core diplomatic portfolios rather than niche ones.