AI chatbots, including the popular ChatGPT, have been found to generate significant factual errors and spread misinformation in the run-up to a recent Scottish election, according to a study by think tank Demos. This raises serious concerns about the integrity of future electoral processes and highlights the need for greater scrutiny of AI technologies.
The research has prompted the Electoral Commission, the body overseeing elections in the UK, to call for new controls and regulations on AI tools that could spread false information and influence public opinion. The potential risks are clear: if AI-powered tools can fabricate scandals, invent non-existent candidates, or provide incorrect election dates, they could have a significant impact on voter decisions.
The Demos study reveals a pattern of inaccuracies in AI responses to queries about the Scottish election, including entirely fictional political controversies and misstated electoral details. These errors are not only disturbing but also pose a novel challenge for electoral bodies tasked with ensuring fair and transparent elections.
The implications extend far beyond Scotland, as the UK prepares for future national and local elections. Political parties rely on accurate information dissemination to engage voters effectively, and AI-generated misinformation could further complicate campaigning efforts and voter engagement. Ensuring that voters receive verifiable information is crucial for informed decision-making at the ballot box.
The UK's Electoral Commission is taking a proactive approach by pushing for new controls, aiming to strike a balance between technological innovation and democratic principles. Discussions around these new controls will involve government departments, technology companies, and civil society organisations, highlighting the need for collaboration to mitigate the risks associated with AI use in politics.