Would you have your personal AI Jonas?

Have you heard of ”AI-Jonas”, a parody of Prime Minister Jonas Gahr Støre in the podcast of the humor duo Tore Sagen and Harald Eia? Thanks to artificial intelligence (AI), his voice is strikingly similar and he speaks credibly about the importance of sustainability in the short, long and medium term. But AI-Jonas is also funny, warm and unpredictable; qualities that don't always show up in Støre's own communication.

What if AI-jonas were not a humorous fixture, but the Labour Party's innovation to make Støre more popular? It wouldn't be the first time a politician would adopt AI to boost his popularity.

Pakistan's former prime minister, Imran Khan, is in jail, but still speaking to the people with the help of AI. As early as 2020, an Indian state candidate used AI to appear fluent in a specific Hindi dialect. The new president of Indonesia used AI to create cartoonish avatars of him to make him gentler and easier to like.

This is just the start. What if politicians get access to personal data that Google and Facebook have collected about us? Then AI Jonas can adapt his message to each of us. Each of us could have had our own bespoke PM in our ear. Could it be more “1984” than that?

Today it is hard to imagine that something like this will happen. AI jonas would probably lose in the competition for attention with other AI influencers. In addition, there are limits to how much personalized content we will accept. A completely customized message can be ineffective. We want to experience the same things as those around us. I want to discuss what Støre says to us, not just what he says to me.

Even if it could work, it is better if AI is used to open a democratic channel, where the people can speak directly to the politicians who represent them.

There is a lot of talk about how AI can threaten democracy, but democracy as we know it is full of limitations.

We have the power to influence politics in Norway by casting a vote every two years. However, the parties we vote for are given no information about why we voted for them or what we want them to do. AI can be used to create a more information-rich channel.

It is, of course, already possible for voters to communicate directly with their elected officials. Politicians' e-mail addresses are available online, one can submit input on party program processes and arrange meetings with parliamentary representatives. But such contact is reserved for a small minority.

It's not just because too few people are interested in politics. Politicians do not have any opportunity to systematize all the information they would have received if citizens showed greater interest. AI is the solution.

The technology can be used to capture large amounts of qualitative information and to compile it qualitatively and quantitatively, as chatGPT can create summaries of larger amounts of text.

What could this have looked like? The authorities could offer every citizen an AI Socrates: your personal political sparring partner. Always curious, always listening.

The AI Socrates would have served as a sophisticated election machine, giving users a better understanding of their political preferences. It would also create a huge amount of information about what voters think.

AI could also have been used to analyze and compile the information of millions of voters, so that politicians pursue policies that are more in line with the wishes of the people.

To avoid people abusing AI Socrates by creating many accounts and thus giving the impression that one's political desires are shared by several, one should require a login with BankID.

Secure, digital certification systems such as BankID can solve a number of the problems with AI. Many are concerned that AI should be used to spread misleading information through newspapers and on social media. But if everyone had to prove who they are, to write in the newspaper and post on social media, this problem would have been solved.

That doesn't solve the problem that comedians can usie AI to generate fake interviews in their own name. But it's not a more recent problem than the fact that one at all times has been able to tell untruths to journalists. We will still have to put our trust that most people who speak out in their own name will think of their good name and reputation.

In order for AI Socrates to function optimally, the information must be shared with the general public. We cannot trust that politicians will act in line with the wishes of the people if the information is not available to the media and interest organisations. However, this could have been resolved if AI Socrates had been governed by an independent agency with a commitment to making the information available to the public and government in a way that is compatible with people's private lives.

In complicated questions that require a lot of prior knowledge, it can be problematic to follow people's wishes directly. But in two areas in particular, AI Socrates can add much needed knowledge. In matters of value, AI Socrates can help politicians understand what is important to the population. In local politics, people know the politics on the body, and politicians will be able to benefit from each individual's knowledge.

AI also provides scope to include citizens in a different way. Social media was intended to bring people together in a larger, public conversation, but does not live up to the ideal. However, the authorities should step in where the private providers fail. Norway should look to Taiwan and other countries that offer platforms for a more constructive conversation. AI can strengthen the democratic function of social media, for example by showing where there is agreement and disagreement on political issues and where there is room for compromise.

Social media and AI technology can strengthen, rather than weaken, democracy. If we use it in the right way.

‍