Despite the pervasive use of artificial intelligence (AI) technologies, many Americans still harbor deep reservations about their reliability and credibility. Notably, Jim Duggan, a traditionalist and dedicated supporter of the Republican party from Huntsville, Alabama, integrates AI into his carbon offset initiative but remains skeptical about its role in the political sphere. He argues that while AI excels in matters of syntax and word choice, it is ill-suited to distill political thought, views, and nuanced judgements.
This skepticism is shared by a majority of the US populace, according to a poll conducted by The Associated Press-NORC Center for Public Affairs Research and USAFacts. The study found a considerable lack of faith in AI-powered chatbots or search results, demonstrating a fundamental distrust in these digital tools’ ability to offer precise and factual information.
With the adoption of AI technology in our daily lives and workplaces, there is a growing suspicion towards these rapidly evolving tools. This skepticism is especially prevalent concerning information related to significant events, such as elections, where the consequences of misinformation can be profound.
In a recent rendezvous of election officials and AI researchers, AI tools were reported to struggle with even basic election-related inquiries, including, for instance, locating the nearest polling place. This insufficiency was underlined when a social media platform’s AI chatbot was accused of sharing spurious election information. This incident led to an urgent modification of the AI tool to redirect users to a federal government website where accurate information is guaranteed.
The complex workings of large AI models that generate various content forms remain largely misunderstood and minimally regulated. While these models deliver refined responses on almost any subject, their predictive capabilities, hinged on extensive data, also make them susceptible to errors. This peculiar duality has put the American public in a quandary.
Americans have diverse views regarding AI’s impact on the ease of accessing accurate election information for the 2024 elections. Some predict a more challenging journey, while others remain undecided, unsure whether AI will simplify or complicate the process. Only a marginal 16% of the population believes AI will facilitate access to correct election information.
A Democrat student from Texas, Griffin Ryan, voices his skepticism towards the use of AI for political information. He notices the ability to manipulate AI responses and thus opts for traditional news outlets. His main concerns are the advent of AI-generated deepfakes and AI bots that may manipulate voter opinions.
Despite being aware of AI deepfakes mostly in the form of harmless humor, Ryan remains wary of their potential malicious use in disseminating deceiving content. He sees the power of deepfake technology as a serious force with possible negative implications.
According to the poll, a tiny percentage, merely 8%, believe AI models like OpenAI’s ChatGPT or Anthropic’s Claude are consistently grounded in factual data. Similarly, AI-assisted search engines like Bing or Google inspire little confidence, with only 12% trusting their results to be fact-based.
There have been documented instances of AI misuse in US elections, including the deceptive use of deepfakes and AI-generated robocalls designed to dampen voter turnout. The more frequent abuses involve creating fabricated images of high-profile candidates to amplify damaging narratives.
Ryan expresses concern regarding the susceptibility of older relatives to misinformation, hinting at the generation gap in the comprehension and critical analysis of digital media as a source of truth. These fears are not baseless, given their previous adherence to false COVID-19 vaccine information encountered on social media platforms.
In contrast, Bevellie Harris, a senior Democrat from Bakersfield, California, relies mainly on official governmental sources for election-related information. She views the traditional voter pamphlet received in the mail preceding each election as the most valuable source of information, offering an example of the enduring trust in official channels.
Recognizing the variety of information sources, Harris also appreciates listening to candidates articulate their platforms firsthand through campaign advertisements. Her approach shows the informed voter’s ability to combine the efficiency of modern communication with traditional channels to ensure balanced and accurate information.