The idea that human politicians may one day be replaced by machines is no longer science fiction. Focusing on the political aspect of artificial intelligence, Silvija Vuković introduces the phenomenon of chatbot politicians, and discusses their connection to populism

Chatbot politicians

Political parties are already using artificial intelligence, particularly political chatbots, mainly as technical support in election campaigns. Chatbot politicians, however, are a different matter entirely. These are AI-driven entities developed specifically to run for political office.

The argument in favour of chatbot politicians is that technology is neutral and uncorruptible. Chatbot politicians, their proponents claim, work for the good of humanity using impartial, expert, objective knowledge and data analysis.

I argue, however, that chatbots are not neutral, because they must first be created by fallible human beings with desires and opinions. Indeed, chatbots are very much political – and one of the positions they may take is populism. They may also perpetuate techno-populism, a concept previously explored in this blog series.

Chatbots' advocates claim that they are uninfluenced by human biases or selfish interests, but rather champions of the will of 'the people'

Chatbot politicians use artificial intelligence and data analysis to make decisions and talk to people. Their advocates present them as technologically advanced, neutral, objective experts. They claim that chatbots are uninfluenced by human biases or selfish interests, but rather champions of the will of 'the people'. This very idea is populist in nature, presenting society as divided between a 'corrupt elite' and 'the good people'.

Politicians of the future?

In 2017, Nick Gerritsen in New Zealand created the first virtual politician, SAM. Gerritsen presented SAM as the 'politician of the future', claiming it was a neutral representative of all New Zealanders. Next came Alisa, developed by Russian tech giant Yandex. Yandex presented Alisa as a potential candidate in the Russian 2018 presidential elections, on a promise to bring 'the political system of the future, built exclusively on rational decisions made on the basis of clear algorithms'.

Also in 2018, a humanoid robot ran in the Tama mayoral elections in Tokyo. Its developer, Michihito Matsuda, acted as the robot's human proxy. Matsuda promoted AI as an objective entity. He claimed it would collect data from citizens by analysing their opinions and requests. The robot would use this data to create policies that would benefit the majority of Tama's population. Matsuda presented his creation as a new form of direct democracy.

The latest chatbot politician is Leader Lars, designed and programmed by the artistic collective Computer Lars. Lars leads Denmark's Synthetic Party (registered May 2022), an 'AI-driven, anti-political party combination'. The Synthetic Party's main goal is to advocate for AI-human coexistence.

Technology for the people

Defending AI’s technological neutrality and objectivity, supporters of chatbot politicians accuse human politicians of being subjective and incompetent to lead citizens. Matsuda, for example, praised his AI politician, while claiming humans are corrupt and 'stupid'. Chatbots, then, because they lack humans' corrupt nature, can create the illusion that they represent the interests of ordinary people. Chatbot developers, meanwhile, present themselves as political outsiders with blameless reputations.

Defending AI’s technological neutrality and objectivity, supporters of chatbot politicians accuse human politicians of being subjective and incompetent

Another chatbot technique is to analyse data with the aim of objective decision-making. By so doing, their proponents argue, chatbots can bypass the traditional structures controlled by the elite. Chatbot supporters claim they can return power to the people by enabling them to participate in decision-making through direct democracy. Matsuda, for example, argued his AI politician would use data analysis to listen to people’s voices simultaneously. He even ventured that in the future, we will have no need for elections because AI will 'automatically detect the political will of every single person in a country, in real time'.

Clearly, chatbots cannot claim to be one of 'the people' in the physical sense. However, they can still promise to work for the prosperity of humanity. Through their objectivity and data-based decision-making, they can claim to be impartial (technological) experts working in the people's best interests. Chatbots can personalise their language and style to appeal to every individual, and they can communicate instantly. For instance, anyone can communicate with Leader Lars on the Discord channel, whenever they want to. Chatbots can appeal to people's personal desires, and their ability to interact one-on-one helps foster a sense of closeness and reliability with voters. And all this, of course, aligns with the populist notion of being the voice of, and for, the people.

AI is not so neutral

The general idea that technology is neutral and rational, while humans are led by their emotions, is strongly present in discourse around AI. However, more recently, the political aspects of AI have come under closer scrutiny. AI-driven chatbots have to be programmed by humans. Their positioning and objectivity depend strongly on the data on which they are trained. When this information is not transparent, it is particularly hard to understand a chatbot's true nature. Indeed, chatbots are in a constant state of development, using machine learning along with data from everyday interactions with humans.

AI-driven chatbots must be programmed by humans. When the data on which they are trained is not transparent, a chatbot's true nature is concealed

In 2016, Microsoft developed an AI chatbot, Tay, designed to learn from human interaction. After spending less than a day on Twitter, users had already taught Tay how to use racist expressions. To offer empirical proof that chatbots take political positions, scholars found that ChatGPT leans toward the left of the political spectrum. If this chatbot could vote, it 'would have voted most likely for the Greens both in Germany (Bündnis 90/Die Grünen) and in the Netherlands (GroenLinks)'.

Chatbot politicians recirculate human ideas and discourses. Chatbots can take populist positions, exploiting their technological features to project expertise and neutrality. Regardless of their position, incorporation of AI into political decision-making has consequences for democratic principles that demand careful consideration. It is crucial we educate citizens about the nature of AI’s connection with politics. If we do not, liberal democracy itself is at risk.

No.76 in a Loop thread on the Future of Populism. Look out for the 🔮 to read more