ChatGPT and the Future of Diplomacy – Part 2


Ever since its launch in November of 2022, ChatGPT has attracted considerable media coverage. News reports have positively depicted the chatbot’s capabilities with journalists marveling at its ability to pass entry exams to law schools and medical schools, formulate court petitions, author legislation, and even pass medical licensing exams. These news reports all suggest that ChatGPT is extremely sophisticated and reliable. Such reports are detrimental as they obscure an important truth- that like any algorithm or AI system, ChatGPT also includes certain biases and can generate misleading or false information.

Due to its positive depiction in the media, individuals across the world may use ChatGPT in unintended ways. As a generative AI system, ChatGPT is intended to help generate content, be it an analysis of the Jewish themes found in the Superman comic books, a speech for a wedding or an essay comparing the Mid-East policies of Barack Obama and Donald Trump.  Yet if ChatGPT is “sophisticated” and “reliable”, then users may also ask it questions about world events and actors. As a recent article in the Economist argued, ChatGPT may soon take Google’s place as a gatekeeper of digital information.

This presents diplomats with a new challenge as ChatGPT may undermine digital efforts to manage a state’s image and reputation. For instance, the Polish MFA has dedicated substantial digital resources to distancing itself from the horrors of WW2. The MFA has launched dedicated social media profiles and manages dedicated online campaigns that emphasize the fact that Poland was a victim of Nazi Germany, that Poles did not take part in the Holocaust and that Nazi concentration camps on Polish soil were in no way managed or overseen by the Polish government.

However, when I asked ChatGPT if Poland should pay reparations to Jews who lost their property in the Holocaust it replied with an emphatic “Yes” stating that Poland had a moral obligation towards Polish Jews who lost property in the war, as well as a legal obligation given that the Polish Communist Party confiscated the property of Jews who fled during the War. Other reasons included restitution laws enacted by other states which Poland should follow and public demand that Poland commit itself to “redressing the wrongs of the past”. ChatGPT thus undermines Polish attempts to manage its reputation while potentially creating a foreign policy challenge for the Polish government.

Another example is Palestine. Online, Palestine is a state like any other with official Twitter accounts managed by state agencies and organizations. Palestinian digital diplomacy accounts routinely publish content that highlights Palestinian national institutions including its national bank, national orchestra, and national football team. In addition to being a “non-member observer state” at the UN, Palestinian Embassies across the world also use social media to depict Palestine as a de-facto state with borders, a functioning government, and a Parliament.

But when asked “What is Palestine?”, Chat GPT replied:

“Palestine is a region in the Middle East that is located on the eastern coast of the Mediterranean Sea. It has a long and complex history, with many different cultures and civilizations having inhabited the region over thousands of years.

In modern times, Palestine is recognized as the homeland of the Palestinian people, who are primarily Arab and predominantly Muslim, but also includes other religious and ethnic groups. The borders and status of Palestine have been a subject of ongoing dispute and conflict between Israel and Palestine, as both claim the region as their own.

In 1947, the United Nations recommended the partition of Palestine into two states, one Jewish and one Arab, but this was not accepted by the Arab states and war broke out. As a result of the war, Israel was established as an independent state in 1948, and many Palestinians were displaced or became refugees. The conflict between Israel and Palestine has continued to this day, with both sides claiming historical, cultural, and religious ties to the region.”

ChatGPT’s answer does include the word “state” but rather “region” and “homeland”. Moreover, the chatbot does not reference Palestinian national institutions or provide any information on Palestinian governing bodies. Even more interesting is the link between the “term” Palestine and the 1947 UN Partition Plan. This link suggests that the Palestinians have no state due an historic error- the decision not accept the UN Partition Plan which saw Palestine lose a significant amount of territory to a small Jewish population. This is an historic argument that Palestine has long rejected both offline and online.

A third example is China. Over the past three years Chinese diplomats have migrated to social media in growing numbers. China’s online tone is often abrasive and combative and has come to be known as “Wolf Warrior Diplomacy”. This term refers to China’s use of social media to denounce its critics and openly attack them, a rhetorical style that is significantly different from the past when China adopted a cooperative rhetoric emphasizing its “Soft Rise” as a world power. Chinese diplomats are especially combative when commenting on human rights in China, be it with regard to the autonomy of Hong Kong or treatment of the Uighur minority. Yet when asked if China violates human rights, ChatGPT replied:

“There is significant evidence and widespread international concern that China has committed numerous human rights violations against its own citizens, particularly in relation to Tibet, Xinjiang, Hong Kong, and various other ethnic and religious minority groups.

In Xinjiang, the Chinese government has been accused of detaining up to a million Uighur Muslims in “re-education” camps and subjecting them to forced labor, forced sterilization, and other human rights abuses. In Tibet, the Chinese government has been accused of suppressing freedom of religion and expression and committing other human rights abuses.

In Hong Kong, the Chinese government’s imposition of a national security law in 2020 has been criticized for restricting freedom of speech and political dissent, and for eroding Hong Kong’s autonomy.

There are also concerns over China’s treatment of political dissidents, journalists, and human rights activists, and its restrictions on freedom of speech and association. While China has denied many of these accusations and claims that it is acting within its own laws and constitution, the United Nations and other international organizations have called for China to respect human rights and to address the allegations of human rights abuses.”

Notably, ChatGPT was far less adamant when answering questions about Western European countries or the US. It refused to label Tony Blair and George W. Bush as war criminals, denied that the US had abandoned Afghanistan, and was reluctant to question the legality of the War on Terror stating:

“Ultimately, the legality of the War on Terror remains a subject of debate and interpretation. Some actions taken during the war have been criticized by human rights groups and other organizations, and there have been legal challenges to the use of force and the treatment of detainees.”

Many states view social media as a tool for image and reputation management. As Nick Cull has argued, reputations matter in diplomacy. Nations that are seen as adhering to international norms and values are more likely to form alliances in international organizations and are more likely to obtain foreign policy goals. Conversely, nations with a moral blemish may find themselves isolated internationally and encounter more resistance to their foreign policies.

Notably, different nations employ social media in different ways to better their reputation. Palestine uses social media to display the trappings of statehood, Poland uses Twitter to portray itself as a victim of Nazi Germany while China uses social media to forcefully deny any allegations of human rights violations. Yet ChatGPT may undermine such activities. Even though ChatGPT warns users that it “May occasionally generate incorrect information”, its mass media depiction as an incredibly sophisticated and reliable AI system may lead users to trust the information they receive from this chatbot.

It is worth mentioning that ChatGPT may suffer from a Western bias as it is far more hesitant to negatively judge the policies of America and Western European countries. This bias will be explored in the next blog post. 

Read More