X, formerly Twitter, has bowed to pressure to add a warning message to its Grok AI chatbot tool that urges users to seek information about the upcoming US presidential election from official sources.
The joint request was made to the social media platform by five Secretaries of State after it was found that Grok was producing false information in response to queries about the election.
The result is that, when asked related queries, the chatbot will now direct users to an official governmental website where they can find “accurate and up-to-date information”.
Ensuring Voters Have Accurate Information
On August 5th, the Secretaries of State for Michigan, Minnesota, New Mexico, Pennsylvania and Washington sent a co-signed letter directly to X owner Elon Musk stating that they and their combined 37 million constituents had been impacted by false information coming from the platform.
Specifically, they cited incorrect information given by the Grok chatbot around ballot deadlines after Joe Biden announced he was withdrawing his presidential candidacy last month.
This just in! View
the top business tech deals for 2024 👨💻
Grok said that the ballot deadline had passed in nine states (including those of the co-signing officials), which was not the case.
The Secretaries of State asserted an anxiety that potential voters should have accurate, up-to-date elections information.
“X has the responsibility to ensure all voters using your platform have access to guidance that reflects true and accurate information about their constitutional right to vote.” – Secretaries of State letter
It requested in conclusion that X would “immediately adopt a policy of directing Grok users to CanIVote.org when asked about elections in the U.S.”
Free, Fair, Secure, and Accurate
As originally reported by the Associated Press, X has now added a warning message that will show before Grok gives any responses to election-related questions:
“For accurate and up-to-date information about the 2024 U.S. Elections, please visit Vote.gov.”
The petitioning officials welcomed the move in a further joint statement, describing Vote.gov as a trustworthy resource.
“We appreciate X’s action to improve their platform and hope they continue to make improvements that will ensure their users have access to accurate information from trusted sources in this critical election year,” the statement reads. “Elections are a team effort, and we need and welcome any partners who are committed to ensuring free, fair, secure, and accurate elections.”
AI and the Risk of Misinformation
In the original letter, the officials acknowledged that the nature of even the best AI chatbots and their large language models meant that there would always be a risk of misinformation.
That risk has already reared its problematic head in election campaigns this year, most notably with last week’s news of false images showing Taylor Swift endorsing Donald Trump. While ahead of June’s UK general election, ChatGPT refused to answer election questions altogether after being found to have given inaccurate answers.
And it’s not just the election. In recent weeks UNESCO issued a warning that AI could spread false information about the holocaust, and a research study showed that ChatGPT gets medical diagnosis wrong half of the time.
We suspect that the already lengthy list of AI errors, mistakes and failures won’t stop growing any time soon.
