HomeArtificial IntelligenceFrom deepfakes to digital candidates: the political game of AI

From deepfakes to digital candidates: the political game of AI

AI is increasingly getting used to represent or misrepresent the opinions of historical and current figures. A recent one Example President Biden's voice was cloned and utilized in a robocall to New Hampshire voters. Going a step further, given AI's advancing capabilities, the symbolic “candidacy” of an AI-created persona could soon be possible. This could seem far-fetched, however the technology to create such a political AI actor already exists.

There are many examples that time to this possibility. Technologies that enable interactive and immersive learning experiences bring historical figures and ideas to life. If used responsibly, these can’t only demystify the past but additionally encourage a greater informed and more engaged citizenry.

People can do it today interact with chatbots They reflect the viewpoints of personalities from Marcus Aurelius to Martin Luther King Jr. with the Hello History app or George Washington and Albert Einstein with Text with History. These apps claim to assist people higher understand historical events or “just have a good time chatting together with your favorite historical characters.”

There can also be a Vincent van Gogh exhibition Musee d'Orsay in Paris features a digital version of the artist and offers viewers the chance to interact with him. Visitors can ask questions and the Vincent chatbot responds based on a training dataset of greater than 800 of his letters. discusses other examples, including an interactive experience in a World War II museum that offers visitors the chance to talk with them AI versions of military veterans.

The worrying rise of deepfakes

Of course, this technology may also be used to clone each historical and current public figures with different intentions and in ways in which raise ethical concerns. I’m referring here to the deepfakes which can be increasingly spreading and making it difficult to differentiate real from fake and truth from falsehood, as mentioned in the instance of the Biden clone.

Deepfake technology uses AI to create or manipulate still images, video and audio content, making it possible to convincingly swap faces, synthesize speech, and fabricate or alter actions in videos. This technology mixes and manipulates data from real images and videos to provide realistic-looking and sounding creations which can be increasingly difficult to differentiate from authentic content.

Although there are legitimate educational and entertainment purposes for these technologies, they’re increasingly getting used for less confident purposes. There are many concerns concerning the potential of AI-generated deepfakes that impersonate well-known figures to govern public opinion and potentially rig elections.

The rise of political deepfakes

Just this month there have been stories about AI getting used for such purposes. Imran Khan, the previous prime minister of Pakistan, effectively campaigned from prison with speeches created using AI to clone his voice. This was effective as Khan's party performed surprisingly well in a recent election.

As written in: “‘I had every confidence that you simply would all vote. “You have fulfilled my trust in you and your tremendous participation has amazed everyone,” said the soft, barely robotic voice within the minute-long video, which used historical images and pictures of Mr. Khan and included a disclaimer about his AI origins.

This was not the one recent example. A political party in Indonesia has created an AI-generated deepfake video of former President Suharto, who died in 2008. In the video, the fake Suharto encourages people to vote for a former army general who was a part of his military-backed regime. As CNN reported, this video, released just weeks before the election, was intended to just do that influence voters. And it did, it got 5 million views. The former general won the election.

Similar tactics are getting used in India. Aljazeera reported that recently an icon of cinema and politics, M. Karunanidhi appeared in front of a live audience on a big projection screen. Karunanidhi delivered a speech by which he “effusively praised the able leadership of MK Stalin, his son and the present head of state”. Karunanidhi died in 2018, but this was the third time within the last six months that he had “performed” for such public events via AI.

It is now clear that the AI-powered deepfake era in politics, first feared several years ago, has fully arrived.

Imagine the rise of “artificial” political candidates

Techniques resembling those utilized in deepfake technology create highly realistic and interactive digital representations of fictional or real characters. These developments make it technologically possible to simulate conversations with historical figures or create realistic digital personas based on their public records, speeches and writings.

A possible recent application is for somebody (or a bunch) to nominate an AI-created digital persona for public office. Specifically, it’s a chatbot that supports AI-created images, audio and video content. “Fancy,” you say? Naturally. Ridiculous? Quite possible. Plausible? Complete. After all, they already function therapists, FriendsAnd girlfriends.

There are several obstacles to this concept, not the least of which is the proven fact that an actual candidate for Congress and even a neighborhood city council have to be an actual person. Therefore, a chatbot cannot register as a candidate or register to vote.

However, what if a write-in campaign resulted in a digital persona chatbot receiving more votes than another candidate on the ballot? This seems implausible, nevertheless it is feasible. Since this is only hypothetical, we are able to play out an imaginary scenario.

Do you could have milk?

For the sake of dialogue, let's assume that “Milkbot” is a candidate for a future San Francisco mayoral election. Milkbot uses an open-source LLM (Large Language Model) based on the writings, speeches, videos and social posts of Harvey Milk, the late former member of the San Francisco Board of Supervisors. The data set may very well be supplemented with content from individuals who had or hold similar views.

Milkbot can provide speeches that its promoters help create, create AI-generated videos and audios and publish them on various social platforms. Milkbot can also be able to “answering” questions from the general public, just like Vincent van Gogh, and as its popularity increases, also answering questions from the press. Because of the novelty or because no real candidate is drawing public attention within the election, momentum is growing for Milkbot's mayoral efforts.

The bot then receives more votes through the write-in campaign than another candidate on the ballot. It is feasible that the vote is symbolic and corresponds to “not one of the above”, nevertheless it is also that the result corresponds to the needs of the voters. What happens then?

The election authorities would probably simply declare the result inadmissible and the human candidate with the best variety of votes could be declared the winner. However, this final result could also result in a legal redefinition of what constitutes a candidate or winner in a political contest. There would definitely be questions on representation, accountability and the potential for manipulation or abuse of AI within the political process. Of course, comparable questions exist already in the actual world.

Last but not least, the potential for using a digital persona in a symbolic campaign could appear as a type of social or political commentary. These bots could highlight topics resembling dissatisfaction with current policy options, the will for reform, exploring futuristic concepts of governance, and sparking discussions concerning the role of technology in society, the character of democracy, and the best way humans interact with AI should.

This possibility will open up further ethical debate. For example, would a “candidate” digital persona entry be an abomination, or if it gained support, would this be a designer democracy by which the candidate can promote certain policies and characteristics?

Imagine a digital person being nominated for even higher office, perhaps on the federal level. When the robot revolution for politicians comes, we are able to hope that the machines might be trained for integrity.


Please enter your comment!
Please enter your name here

Must Read