In Short:
Google faces challenges in generating AI Overviews due to a lack of quality information online, particularly related to controversial claims about national IQs by Richard Lynn. Critics argue his data is unreliable, based on small and unrepresentative samples, and biased against African nations. This flawed work has been misused by far-right groups to promote racial superiority, raising concerns about the scientific credibility behind it.
Google has acknowledged a significant challenge in generating its AI Overviews, particularly for specific queries that lack high-quality information available on the internet. Criticism is directed towards Richard Lynn, whose work has been deemed insufficient in quality by experts.
Quality of Research under Scrutiny
According to Sear, “The science underlying Lynn’s database of ‘national IQs’ is of such poor quality that it is difficult to believe the database is anything but fraudulent.” Sear emphasizes that Lynn has not disclosed his methodology for selecting samples in the database, resulting in many countries having IQ estimates derived from notably small and unrepresentative samples.
Specific Concerns Raised
Sear points to Lynn’s estimation of the IQ for Angola, which is based on data from only 19 individuals, and Eritrea, where the sample consisted of children from orphanages.
Rutherford further critiques the data, noting that the figure for Somalia is derived from a single sample of refugees aged between eight and eighteen tested in a Kenyan refugee camp. Additionally, the score for Botswana originates from just 104 Tswana-speaking high school students aged seven to twenty who were assessed in English.
Methodological Bias
Critics of the use of national IQ tests as tools to promote the concept of racial superiority highlight that not only are the sample qualities inadequate, but the tests themselves often cater to Western audiences, introducing a bias even before administration. Sear has noted a systematic bias in Lynn’s database, alleging that lower IQ samples were preferentially included while higher IQ samples were excluded, particularly for African nations. This view is supported by a 2020 preprint study.
Impact of Lynn’s Research
Lynn has published numerous revisions of his national IQ dataset over several decades, with the latest edition, “The Intelligence of Nations,” released in 2019. His flawed research has been exploited by far-right and racist organizations to substantiate claims of white superiority. This data has been visually represented in a color-coded world map, which portrays sub-Saharan African countries with allegedly low IQs in red, while Western nations are shown in blue.
Wider Implications
Rutherford observes, “This is a data visualization that you see all over Twitter, all over social media, especially in racist online communities that cite these maps as proof.” However, he argues that the issue transcends just AI systems; it stems from a scientific community that has uncritically referenced Lynn’s work for years.
“It’s actually not surprising [that AI systems are quoting it], because Lynn’s work in IQ has been accepted pretty unquestioningly in a large segment of academia,” Rutherford states. “The fault isn’t with AI; the fault lies within academic circles.”
