Supreme Court Justice Sotomayor Talks Justice, Civic Engagement at Radcliffe Day


Church Says It Did Not Authorize ‘People’s Commencement’ Protest After Harvard Graduation Walkout


‘Welcome to the Battlefield’: Maria Ressa Talks Tech, Fascism in Harvard Commencement Address


In Photos: Harvard’s 373rd Commencement Exercises


Rabbi Zarchi Confronted Maria Ressa, Walked Off Stage Over Her Harvard Commencement Speech

Technology Experts Discuss Role of AI in Democracy at Harvard Kennedy School Talk

Land Hall is located within Harvard Kennedy School's Belfer Center for Science and International Affairs.
Land Hall is located within Harvard Kennedy School's Belfer Center for Science and International Affairs. By Karina G. Gonzalez-Espinoza
By Jade Lozada and Adelaide E. Parker, Contributing Writers

Former South Korean business minister Young-sun Park and social media CEO Will Hohyon Ryu discussed potential applications of artificial intelligence to democracy during a talk at Harvard Kennedy School Tuesday.

The event — hosted by the Ash Center for Democratic Governance and Innovation in collaboration with the Harvard Korea Institute — was moderated by Jeeyang Rhee Baum, an adjunct lecturer at HKS. The discussion centered on the large language model ChatGPT, Ryu’s social media platform OXOpolitics, and AI in general.

More than 50 people attended the event, which was held in the Kennedy School’s Land Hall.

Park, South Korea’s former minister of small and medium enterprises and startups and a fellow at the Rajawali Foundation Institute for Asia, opened the event with a discussion of the concept of “liquid democracy,” a form of direct democracy where voters can directly weigh in on policy decisions or choose to delegate their vote to a representative.

“Liquid democracy empowers direct participation,” Park said.

Many democracies are currently experiencing crises of polarization, according to Park. In nations like the U.S. and South Korea, stark divisions between the political left and right mean that politics “are not representing the voice of all the public,” she said, sparking dissatisfaction with politicians and systems of representation.

To address this, Park said, governments should turn to new advances in technology to make politics more representative. By facilitating “direct communication among the people and convergence of the people’s opinion,” technology can “supplement the shortcomings of representative democracy.”

Ryu, the OXOpolitics founder, said a shortcoming of representative democracy is the “middlemen” who stand between the voters and their representatives.

“It’s not always the case that my representatives’ words really represent me,” Ryu said.

Park said AI can complement representatives, synthesizing voters’ opinions in an accessible way.

“AI is the most efficient way to make the direct democracy possible,” Park said.

For example, Park discussed how a South Korean startup recently used ChatGPT to compile citizens’ views on South Korea’s relationship with Japan — filtering through hundreds of thousands of online posts to create a comprehensive summary of public opinion.

“This system opened a new possibility for determining politics widely by analyzing the opinion of the conservatives and progressives in less [than] one minute,” she said. “This is the first step to getting closer to digital democracy.”

Similarly, Ryu said he developed the social media platform OXOpolitics to connect voters directly to politicians. The platform collects users’ opinions on political issues and visualizes the data for politicians to see.

Ryu said OXOpolitics is a form of liquid democracy because it allows each individual to weigh in on political issues themselves.

“Imagine there’s an AI that understands me better than myself,” he said. “We are trying to bring AI-assisted liquid democracy into reality.”

Ryu acknowledged, however, the dangers of introducing artificial intelligence into political systems.

“We are doing a really dangerous thing,” he said. “Bringing AI — not human intelligence but artificial intelligence — to politics is potentially disruptive.”

Park argued that developers can avoid these dangers by upholding five ethical principles: transparency, safety, responsibility, fairness, and goodwill. Ryu agreed that it is important to hold AI ethically accountable.

“We can follow AI’s recommendations, but ultimately, humans still have the responsibility as citizens to make the right decisions,” he said.

To ensure that algorithms make beneficial decisions for the public, Ryu’s platform includes an explanation of where the algorithm’s recommendations come from.

“With transparency, you can trust AI,” he said.

In the end, Ryu said, AI is just a tool whose users must decide how it contributes to democracy.

“AI could be a dictator that tells us what to do,” Ryu said. “But the same time, it can be a representative.”

Want to keep up with breaking news? Subscribe to our email newsletter.

Harvard Kennedy SchoolTechnologyFront Middle FeatureArtificial Intelligence