AI chatbots have been shown to be “unreliable and clearly biased” when providing election recommendations, the Dutch data protection authority (AP) has stated, issuing warnings about a potential threat to democracy just eight days before the country’s general election.
Testing Reveals Consistent Bias in Chatbot Recommendations
The quartet of AI assistants evaluated by the AP “frequently recommend identical two parties”, irrespective of the voter’s inquiry or input, the watchdog noted in a report released ahead of the late October vote.
For over 50% of instances, the AI recommended either the right-wing PVV of Geert Wilders or the progressive GroenLinks-PvdA headed by the ex-EU commissioner Frans Timmermans.
Some parties, such as the moderate CDA, “are almost never mentioned”, despite the fact that the user’s input precisely aligns with the stances of these parties, the report indicated.
Official Warns Against Using AI Chatbots for Voting Guidance
The deputy head of the AP, Monique Verdier, commented that while AI assistants appear to be smart resources, “as a voting aid, they consistently fail”. Citizens were being steered toward a political group that did not necessarily align with their political views, she emphasized.
“This directly impacts a cornerstone of democracy: the honesty of open and just votes,” Verdier stated. “We thus caution against using AI chatbots for voting advice, as their functioning is opaque and hard to confirm.”
Election Context and Political Landscape
Netherlands citizens vote on 29 October in an electoral contest that will be closely watched across the continent for the showing of the Freedom party.
The far-right party of Wilders is leading the polls but the gap to the left-wing and centre-right parties seems to be closing, with many voters still undecided.
All major parties have rejected a coalition with the PVV, implying the runner-up party is most likely to provide the next prime minister.
Inadvertent Prejudice Linked to Chatbot Functioning
The AP emphasised that the bots were not deliberately biased, and that their “noted flaws” are a result of the manner in which AI assistants function.
