The Ankara Center for Crisis and Political Studies (ANKASAM) presents its interview with Takeo Harada, CEO of the International Institute for Strategy and Information Analysis (IISIA) and Global Artificial Intelligence Strategist, to evaluate the ethical and security risks of artificial intelligence in diplomatic processes, the management of linguistic and cultural differences in diplomatic correspondence and communication, and the impact of AI on the principles of neutrality and justice in diplomatic negotiations.
1. Based on your experiences, what are the ethical and security concerns associated with the use of artificial intelligence in diplomatic processes?
Like in other essential fields of bureaucracy, it is worthy of intensive discussion to apply artificial intelligence to diplomacy, so as to maximize efficiency of labor force working on it. Diplomacy is the art of language as well as human communication, while it is based on responsibility of those who execute it. In this regard, applying AI to diplomacy theoretically raises the following concerns of both ethics and security:
As for the ethics concern, every single data used for AI shall be carefully checked from the perspectives of the ethics in advance, which are diplomatically formulated in the global community so far. Only based on the appropriate data, the outcome of AI can be useful for diplomats, who are professionally obliged to use “correct diplomatic language” for negotiation, documentation as well as public relations. In terms of the security concern, the diplomats making use of AI for their daily desk work shall be extremely careful of the risk that it can be easily manipulated from the outsiders and even enemies through the internet.
Even though the AI is not directly connected to the internet, they have to make sure that it is not pre-trained with inappropriate data. Once it is installed in the diplomacy of a certain country, a large language model (LLM) can be a good tool for propaganda for the sake of other countries, whenever they intentionally pre-trained it with biased corpus of vectorized dataset. In addition, “hallucination” can cause severe trouble in diplomacy, while steadily using LLM-based system. The point is that we can’t technologically avoid it, and we have to check it thoroughly with human evaluation, if we really want to apply it to diplomacy without causing any troubles.
2. In your opinion, how do AI-supported systems in diplomatic correspondence and communication play a role in managing linguistic and cultural differences?
The recent dramatic development of LLM enables us to easily overcome difficulties in communication in general due to differences of both language and culture, which is also the case for diplomacy. Although “English” is used as lingua franca in the modern diplomacy, there are a lot of occasions where local languages of each nation are used. For this reason, diplomats particularly talented for foreign language are required traditionally in diplomacy.
However, the situation changes dramatically, since the level of translation the most modern AI algorithm can achieve is now quite high. In this regard, efficiency of diplomatic correspondence and communication is reaching much higher standard than ever. Nevertheless, the need for “human” diplomat will remain, because both “keeping silence “ and “ speaking out a word with inappropriate phrase” are sometimes of very importance in diplomacy, which AI can never do adequately. In sum, diplomacy will keep on being “human”, but can be automated to some extent thanks to the rapid development and application of artificial intelligence.
3. Do you think artificial intelligence has an impact on the principles of neutrality and fairness in diplomatic negotiations?
In diplomacy, a nation is sometimes asked to play a role as mediator between other countries, for which the principles of neutrality and fairness is strictly needed. In this regard, there is unfortunately little room for AI to be active as well as useful, I think. Artificial intelligence is extremely good at gathering data and information, and sorting as well as classifying them for decision making.
However, to lead any diplomatic negotiations to success for all the stake holders, the mediator is inevitably required to make some “value judgements “ at the final stage of the negotiations. Such “value judgements” can be done only by “human” diplomats, while artificial intelligence is capable of supporting them with precise outcomes of computing just as the basement of the judgements. So far, diplomacy will remain “human” by nature, however better computer power will become in the future, as we see right now in the global realities.
