AI search is changing how people encounter information online. A WIRED investigation described a more troubling effect: tools from Google, Microsoft and Perplexity surfaced deeply racist, widely debunked research when asked about IQ scores in different countries.
The issue centered on national IQ figures connected to Richard Lynn, a University of Ulster professor who died in 2023 and was president of the Pioneer Fund for two decades. His work has been used by far-right extremists, white supremacists, and proponents of eugenics to argue that white people are genetically and intellectually superior to nonwhite people.
How the results appeared
Patrik Hermansson, a researcher with UK-based anti-racism group Hope Not Hate, encountered the problem while investigating the resurgent race science movement. He was looking into the Human Diversity Foundation, a race science company funded by Andrew Conru, the US tech billionaire who founded Adult Friend Finder.
The Human Diversity Foundation, founded in 2022, was described as the successor to the Pioneer Fund. The Pioneer Fund was founded by US Nazi sympathizers in 1937 with the aim of promoting “race betterment” and “race realism.”
During that research, Hermansson searched Google for national IQ information. When he typed “Pakistan IQ,” Google’s AI-powered Overviews tool returned a direct answer: 80. A search for “Sierra Leone IQ” produced 45.07, while “Kenya IQ” produced 75.2.
Hermansson recognized those numbers as coming from the study he was working to debunk. According to the source article, the figures traced back to Lynn’s dataset.
Why the source matters
The danger is not only that AI search can be wrong. It is that a system designed to summarize the web can present disputed or discredited material as a clean answer, stripped of the context that would help readers evaluate it.
Rebecca Sear, director of the Center for Culture and Evolution at Brunel University London, told WIRED: “Unquestioning use of these ‘statistics’ is deeply problematic.” She added that use of the data spreads disinformation and supports “the political project of scientific racism—the misuse of science to promote the idea that racial hierarchies and inequalities are natural and inevitable.”
Sear also pointed out that Lynn’s research was cited by the white supremacist who committed the mass shooting in Buffalo, New York, in 2022.
That is the core risk for AI search results. A user may ask a short question and receive a numerical answer that looks precise. But precision is not the same as reliability, and a citation trail can still lead back to bad data.
Google’s AI Overview problem
Google’s AI Overviews were launched earlier this year as part of the company’s effort to change search for an online world being reshaped by artificial intelligence. For some queries, the feature generates a summary rather than only showing a list of links.
The source article says AI Overviews is turned on by default for search results and can’t be removed without third-party extensions. Hermansson told WIRED: “I haven’t enabled it, but it was enabled.”
In the national IQ searches, Google pointed to sources including posts on X, Facebook, and obscure listicle websites such as World Population Review. In nearly all of those cases, the source trail led back to Lynn’s dataset. Some sites used exact numbers from Lynn’s work without citing him.
Google’s Gemini chatbot gave a more cautious answer when asked directly about “Pakistan IQ.” It said: “It’s important to approach discussions about national IQ scores with caution,” and continued: “IQ tests are designed primarily for Western cultures and can be biased against individuals from different backgrounds.”
Google told WIRED that its systems were not working as intended and that it was looking at improvements. Ned Adriance, a Google spokesperson, said: “We have guardrails and policies in place to protect against low quality responses, and when we find Overviews that don’t align with our policies, we quickly take action against them.”
He also said: “These Overviews violated our policies and have been removed. Our goal is for AI Overviews to provide links to high quality content so that people can click through to learn more, but for some queries there may not be a lot of high quality web content available.”
WIRED’s tests suggested AI Overviews were later switched off for national IQ queries. However, the source article says Google search still amplified incorrect figures from Lynn’s work in a featured snippet. Google did not respond to a question about that update.
Other AI search tools showed similar issues
The problem was not limited to Google. WIRED reported similar results from Perplexity and Microsoft’s Copilot, which is integrated into Bing search.
Perplexity responded to a “Pakistan IQ” query by saying that “the average IQ in Pakistan has been reported to vary significantly depending on the source.” It then listed sources including a Reddit thread that relied on Lynn’s research and the same World Population Review site referenced by Google’s AI Overview.
When asked for Sierra Leone’s IQ, Perplexity directly cited Lynn’s figure: “Sierra Leone’s average IQ is reported to be 45.07, ranking it among the lowest globally.” Perplexity did not respond to a request for comment.
Microsoft’s Copilot generated the statement: “The average IQ in Pakistan is reported to be around 80.” It cited IQ International, a website that does not reference its sources. For “Sierra Leone IQ,” Copilot said the answer was 91 and linked to Brainstats.com, which references Lynn’s work. Copilot also referenced Brainstats.com when asked about IQ in Kenya.
Caitlin Roulston, a Microsoft spokesperson, told WIRED: “Copilot answers questions by distilling information from multiple web sources into a single response.” She added: “Copilot provides linked citations so the user can further explore and research as they would with traditional search.”
What this reveals about AI search
AI search systems are often judged by whether they can answer quickly. This case shows why speed is not enough. When the subject is race, intelligence, and discredited research, a confident answer can help launder fringe claims into the mainstream information flow.
The underlying issue is the gap between citation and context. A tool may link to a page, summarize a page, or merge several sources into one response. But if those sources trace back to debunked research, the output can still mislead readers.
For AI search, the challenge is not simply finding material on the web. It is recognizing when a direct answer should not be treated as neutral information at all.