In 2023, a video surfaced featuring Hillary Clinton endorsing Ron DeSantis, a Republican presidential hopeful. However, investigations revealed that the video was a product of generative artificial intelligence (AI), raising concerns about the potential impact of AI on politics.
Generative AI, as exemplified by the Clinton video, is poised to reshape politics in the near future. Experts warn of consequences, including the creation of false information at little or no cost and the manipulation of voters through highly personalized advertising. This could lead to "October surprises" and the spread of misleading information about electoral administration.
As we approach 2024, a year with numerous elections globally, the use of generative AI in politics becomes a pressing concern. Elections in Taiwan, India, Russia, South Africa, Mexico, Iran, Pakistan, Indonesia, the European Union, the US, and the UK may be influenced by these emerging technologies, paralleling the impact social media had on elections in the 2010s.
While politicians utilized social media to shape elections in the previous decade, generative AI now allows the production of misleading information at virtually no cost. This is alarming, given the role of misinformation, or "bullshit," in politics during the 2010s. With the rise of generative AI, a new phenomenon dubbed "botshit" may emerge, as it effectively reduces the cost of producing deceptive information to zero.
A recent paper by Tim Hannigan, Ian McCarthy, and others explores the concept of botshit. Generative AI, such as ChatGPT, can produce "hallucinations" – answers based on statistically informed guesses, sometimes wildly inaccurate. The risks arise when these outputs have significant consequences, making it difficult to verify the information's accuracy.
If AI-produced hallucinations are used to answer crucial but challenging questions, voters may make decisions based on an illusory information landscape. This raises the risk of voters living in online realities shaped by a mix of AI hallucinations and political expediency.
While acknowledging the dangers posed by AI technologies, there are potential measures to limit their negative impact. Technology companies can use watermarking for easy identification of AI-generated content and ensure AIs are trained on authoritative information sources. Journalists can take precautions to avoid covering AI-generated stories during election cycles, and political parties can develop policies against deceptive AI-generated information. Most importantly, voters must exercise critical judgment by fact-checking important information.
Generative AI has already begun transforming various professions and industries. Politics, according to the Brookings Institution, could experience both positive and negative impacts. It is crucial to ensure that generative AI is used for beneficial purposes, preventing it from contributing to the proliferation of botshit in political discourse.
