Can tech outsmart human thinking?
Modern history is replete with instances of human judgement influencing significant diplomatic occasions, from John F. Kennedy and Nikita Khrushchev negotiating the Cuban Missile Crisis in 1962 to Henry Kissinger and Zhou Enlai’s conversations that prepared the way for China’s opening up in 1971. What if artificial intelligence could assist us in addressing some of the most urgent crises we face today more effectively? Singapore is providing a preliminary insight into what that future could resemble. Last month, the foreign minister of the city-state, Vivian Balakrishnan, disclosed the development of a diplomatic second brain utilising an open-source technology known as NanoClaw. The tool organises transcripts, speeches, and various materials — especially his own contributions — into a searchable database. It has proven to be incredibly beneficial, the minister quipped, that he hesitates to turn it off. The concept embodies a distinctly Singaporean efficiency, tinged with elements of dystopia. Nonetheless, no seasoned diplomat — Balakrishnan included — would wager on an AI system successfully negotiating an end to the conflict in Ukraine or addressing the long-standing tensions between the US and Iran. The minister emphasises that although technology is becoming more capable of managing computation, memory, and routine tasks, it cannot substitute for genuine understanding. Or, as he stated: “You can delegate work, but you cannot delegate accountability.”
In a landscape where diplomatic efforts frequently appear lacking, artificial intelligence has the potential to assist policymakers in navigating intricate challenges. Governments worldwide are currently leveraging AI to analyse vast datasets, simulate various negotiation scenarios, and pinpoint possible areas for compromise. The US State Department is utilising these tools for document translation and information summarisation. In Libya, the United Nations utilised a platform known as Remesh to collect and analyse the perspectives of thousands of citizens during the peace process in late 2020, which came after years of civil conflict. Due to its functionality on basic mobile phones and its support for local dialects, it facilitated a broader range of voices in the discussion compared to what traditional consultations could have achieved. The World Bank is employing AI to forecast refugee movements from South Sudan and the Democratic Republic of Congo into Uganda. This potentially provides governments and aid agencies with additional time to organise housing, schools, healthcare, and other critical services prior to the arrival of individuals.
Despite its potential, AI carries significant risks, as highlighted by Asha Hemrajani. “Garbage in, garbage out,” she stated. “It depends on the data you input into the AI model you are building. And these systems are vulnerable to hacking and manipulation, which could lead to strategic miscalculations. In a high-stakes environment like diplomacy, you have to get it right — you have to keep humans in the loop.” As remarkable as Balakrishnan’s second brain appears, it prompts a thought-provoking inquiry: Whose cognitive function is it supplanting? The tasks that his tool could execute — drafting briefs, synthesising country reports, preparing speeches, and formulating parliamentary questions — have historically been the responsibilities of junior diplomats or policy staff. Diplomacy is not alone in confronting this challenge, yet the issue is particularly pressing in Singapore, a nation that stands out as one of the economies most vulnerable to AI. Economics estimates that approximately 40 percent of employment may be impacted by the technology — the highest proportion globally. Sweden and the UK are positioned at approximately 30 percent, whereas the US is at 26 percent. Exposure does not inherently imply substitution. The figures provide limited insight into the pace of technology adoption or the potential for job augmentation versus elimination. Singaporeans exhibit a notable level of optimism.
Only 17 percent of respondents believe that workers are unlikely to benefit from AI, in contrast to 46 percent in Germany. That could clarify why the government has been able to advance its AI objectives. Singapore has entered into an agreement with OpenAI and, just last week, finalised another with Anthropic. It has also initiated a collaborative AI alliance with South Korea, supported by a $300 million fund, and revealed intentions to implement AI agents throughout the public sector. However, rapid progress carries inherent risks. The challenge lies not in the decision to incorporate AI into diplomacy, but rather in determining the most effective approach to do so. Junior diplomats must cultivate a level of judgement that is beyond the reach of any algorithm. Investing time in language acquisition, cultural comprehension, and gaining insights from seasoned diplomats is crucial. Essential human traits like empathy and understanding play a pivotal role in achieving favourable diplomatic results. The next significant advancement will still necessitate an individual capable of assessing the environment and discerning the cues that could determine the success or failure of a peace agreement. No artificial intelligence-driven second brain can accomplish that.
