AI chatbots are proving to be “unreliable and clearly biased” when providing election recommendations, the Dutch data protection authority (AP) has stated, issuing warnings about a possible risk to democratic processes just eight days before the country’s general election.
Study Shows Repeated Partiality in AI Suggestions
The four chatbots tested by the AP “often end up with the same two parties”, irrespective of the user’s question or command, the regulator noted in a analysis published ahead of the 29 October election.
For over 50% of instances, the AI recommended either the right-wing PVV of Geert Wilders or the leftwing GroenLinks-PvdA led by the ex-EU commissioner Frans Timmermans.
Some parties, such as the centre-right CDA, “are rarely ever cited”, even when the voter’s query precisely aligns with the stances of these parties, the study indicated.
Authority Figure Cautions Against Relying on AI Assistants for Election Advice
The vice-chair of the data protection agency, Monique Verdier, commented that while chatbots may seem like clever tools, “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 added.
“This immediately affects a cornerstone of democracy: the integrity of free and fair elections,” said Verdier. “We therefore warn against using AI chatbots for election recommendations, as their operation is unclear and hard to confirm.”
Election Context and Party Dynamics
Netherlands citizens vote on 29 October in an election that will be keenly observed around Europe for the performance of the PVV.
The right-leaning group of Geert Wilders is leading the polls but the lead over the GroenLinks-PvdA and CDA appears to be narrowing, with many voters still undecided.
All major parties have rejected a coalition with the Freedom party, meaning the party that comes second is highly probable to provide the next prime minister.
Unintentional Bias Attributed to Chatbot Functioning
The authority stressed that the bots were not deliberately biased, and that their “noted flaws” are a consequence of the manner in which AI assistants function.
