When Americans head to the polls in November, we will be part of the first major election with widespread deep-fake capability.
Deep-fakes—essentially, AI-generated, realistic videos and audio files—are more sophisticated and prevalent than ever before. The advances we are seeing will pose significant challenges to electoral integrity—and, in the meantime, are creating a technological race between those seeking to manipulate information and those working to preserve democratic values.
So what does this mean for voters? Thanks to neural network technology, those determined to spread misinformation can create highly convincing fake content at an unprecedented pace, making it harder for us to separate legitimate information from manipulative propaganda. Particularly powerful and hard to combat is visual misinformation, in the form of realistic-looking photos and videos. As improvements in AI continue, it is heating up the ongoing arms race between creators and detectors of fake content.
Addressing AI-driven misinformation will require a combination of organizational and political reforms, as well as technological advancements. Fact-checking efforts continue to evolve and improve but social media platforms, journalists and others involved in rooting out deep fakes struggle to keep pace with AI-driven misinformation. Some social media companies are using manual efforts, others automation, but these efforts are imperfect at best. And as the amount of online content becomes increasingly massive, scaling up fact-checking to keep pace will become an even more difficult task. There is ongoing research into AI-powered detection of false information that is likely to help in the future, but at the moment we have to rely on the methods available to us.
Despite all of this, the greater use of AI does not mean democracy is in decline. AI could potentially bring improvements.
Experiments with decentralized autonomous organizations (DAOs) and deliberative democracy may offer new models for civic engagement. Cryptocurrency and blockchain communities are already experimenting with new governance models that may be relevant. The models allow for decentralized decision-making and voting. If these methods prove effective, they could potentially be adapted to create smaller-scale democratic communities. As an additional benefit, they may offer new ways to foster deliberation and reduce polarization.
Among the experiments in deliberative democracy is research exploring the use of small-group discussions to increase empathy and understanding. So far, these experiments have generated mixed results in fostering long-term attitudinal or behavioral changes. However, if we do uncover more effective methods of deliberation, we could potentially integrate them into electoral processes. The challenge, of course, would be scaling them to large populations.
Another interesting area of investigation is in the use of AI “co-pilots” to potentially assist civic discussions and decision-making. They could bring opportunities to make complex issues more accessible to voters. However, we would need to ensure that AI augments human judgment and doesn’t replace it.
The jury is still out on many of the latest advances and what they portend for democracy. In the meantime, the American public will be part of one of the most important experiments in U.S. history, one that is being conducted by many parties with conflicting agendas but the shared goal of swaying public opinion—and the presidential election.
Learn how Columbia Business School is using AI in the classroom:
