In a 2020 experiment conducted by Cornell University, researchers aimed to assess the ability of elected officials to distinguish between emails composed by artificial intelligence (AI) and those written by human undergraduates. Over 32,000 AI-generated emails were dispatched to state legislators, and the results revealed that the legislators responded to them just 2% less frequently compared to emails authored by humans. This study underscored the remarkable capacity of AI to assimilate extensive human-generated data and autonomously generate information. Nonetheless, it also raised apprehensions regarding the potential exploitation of AI in large-scale influence operations, capable of manipulating politics and disseminating misinformation.
The Danger of Inauthentic Influence Operations
AI-powered influence operations, commonly referred to as “astroturfing,” have a longstanding history in American politics. Previously, these operations required manual efforts, but with the increasing capabilities of AI and computing power, well-funded political entities now have access to sophisticated AI tools. A notable instance occurred in 2017 when the Federal Communications Commission was inundated with a barrage of messages supporting the repeal of net neutrality. Many of these messages were suspected to be generated by AI-powered bots. This incident shed light on the potential of AI to artificially influence public sentiment and deceive policymakers.
Another compelling case highlighting the manipulative power of AI occurred in 2019 when a college student created an AI bot that generated over half of the comments received during a public feedback process in Idaho. This incident exemplifies how AI can be harnessed to fabricate a substantial volume of engagement, skewing public opinion and distorting the perceived consensus on an issue. These examples serve as alarming reminders of the ability of AI to shape public discourse and sway decision-making processes. It underscores the need for vigilance and effective countermeasures to detect and mitigate the adverse effects of AI-driven manipulation in political arenas.
The Rise of AI in Manipulating Politics
AI-driven chatbots have demonstrated remarkable proficiency in emulating diverse forms of human-written text, ranging from professional journalism to conspiracy theories. Extensive research has indicated that AI models, such as OpenAI’s GPT-3 and GPT-4, possess the ability to craft tweets and news articles that sound remarkably authentic. Moreover, AI can function across a wide array of online platforms, even venturing into the depths of the internet’s obscure corners. For instance, the CyberAI Project conducted by Georgetown University discovered that AI models were capable of generating social media posts with the intention of promoting divisive topics and influencing voter turnout. Remarkably, these AI-generated messages often bore no distinguishable difference from content produced by actual human beings.
The Democracy of Propaganda
With the rise of AI in generating political content, concerns have been raised about the democratization of troll farms and content farms. AI tools can empower bad actors and misinformers by providing them with the capability to produce large volumes of content easily. Previously, these actors had to hire individuals to produce such content. This democratization of disinformation poses a significant challenge, requiring vigilant fact-checking and skepticism from users to counter the spread of AI-generated propaganda.
Unveiling AI-Generated Text
Discovering AI-generated text can be challenging, as AI models become increasingly sophisticated. In some cases, human observers have noticed patterns or inconsistencies that revealed the use of AI. However, the ability to detect AI-generated content is not foolproof, and it can lead to confusion and misinformation. This can create a “liar’s dividend,” where real and damaging information is dismissed as synthetic or manipulated. The proliferation of AI-generated content may erode trust in democratic systems if people stop believing anything they encounter online.
The Profit Motive and Skepticism
Skeptics raise an important concern regarding AI, arguing that the profit motive is the real issue rather than political disinformation. AI has already been misused by scammers and criminals to fake kidnappings, create malware, and engage in various forms of abuse. While generative AI can potentially reduce the cost of disseminating disinformation, the primary challenge lies in its distribution. Currently, reaching a wide audience with manipulated content requires significant resources and coordination. However, even a slight reduction in dissemination costs could incentivize bad actors to utilize AI for political gains. To counter disinformation, experts recommend fact-checking and seeking information from multiple trusted sources.
Fact-checking and relying on multiple trusted sources of information are crucial in combating the spread of disinformation. By critically evaluating the content we encounter and verifying its accuracy through reputable sources, we can navigate the complex landscape of AI-generated disinformation more effectively. Additionally, initiatives promoting media and digital literacy play a vital role in equipping individuals with the skills to discern reliable information from manipulative narratives. By empowering people to be critical consumers of information, we can collectively address the challenges posed by AI-driven disinformation and foster a more informed and resilient society.
Living with AI and the Call for Regulation
The coexistence of humans with AI has sparked extensive debate. Skeptics argue that humans will adapt to AI and harness its potential for the greater good, similar to past technological advancements. They believe society will overcome challenges and benefit from AI’s capabilities. However, voices like OpenAI CEO Sam Altman express concerns about AI’s potential for manipulation and disinformation. Altman calls for governmental regulation and responsible corporate stewardship of AI.
AI’s ability to convincingly mimic human text, including journalism and social media posts, raises alarms about its misuse. Experts argue that regulations and responsible practices are necessary to ensure ethical and responsible use of AI. OpenAI recognizes these concerns and supports initiatives to address ethical challenges. They advocate for transparency and accountability in AI development and deployment. Striking a balance between embracing AI’s potential and safeguarding against its potential harms is a crucial task for governments, tech companies, and society.