Next year Americans will go to the polls to choose their next president. For many, the race has already begun. Digital electioneering in US elections has been steadily developing over the last 15 years, but this time around, advancements in artificial intelligence could be a game changer.

There have been huge strides in generative AI in the past year. One of the most accessible AI tools now available to the general public is the software known as ChatGPT, which can scour the internet for information, producing text for speeches and essays. Generative AI is widely used to produce social content around image and text, but what will happen when full on AI video becomes more readily available to any user?

AI systems will be able to reach voters with messages targeted specifically to them, but will they be able to trust them? There are concerns that voters will have an increasingly tough task working out which campaign messages are genuine and which are not. To date, there is currently little regulation of a system which has already been used to create deep-fake manipulations of people and what they say, provoking questions over authenticity.

So do we all have to be more aware of how much we allow AI to shape our democracies?

This week on the Inquiry, we’re asking: Will AI decide America’s next president?

Contributors Betsy Hoover, Higher Ground Labs Prof Hany Farid, University of California Berkeley Martin Kurucz, CEO, Sterling Data Company Nina Schick, author of ‘Deepfakes’

Presented by Tanya Beckett Produced by Jill Collins Researcher: Anoushka Mutanda-Dougherty Editor: Tara McDermott Technical producer: Richard Hannaford Broadcast coordinator: Brenda Brown

Image: Unused privacy booths are seen at a voting site in Tripp Commons inside the Memorial Union building on the University of Wisconsin-Madison campus in Madison, Dane County, Wisconsin, November 3, 2020 (Credit: Bing Guan/Reuters)