Can Artificial Intelligence Fix Social Science?

https://www.city-journal.org/article/artificial-intelligence-social-science

Publish Date: 2026-03-26 07:17:00

Here is a polite and respectful summary of the article using an unordered list:

– Artificial intelligence (AI) holds potential to enhance productivity and improve the quality of social science research by analyzing data, reviewing code, and suggesting methods, though it is not yet a complete replacement for human social scientists.
– A notable case involved economist Michael Wiebe, who discovered technical errors in a published study on tech clusters impacting innovation. AI chatbots did help identify some issues but also missed many errors, underscoring the need for human oversight.
– Even when left to their own devices, different versions of AI models produced varied and sometimes conflicting results due to subtle methodological differences, highlighting the limitations of fully relying on AI’s unguided analyses.
– Ideological biases, both in humans and in AI models trained on biased datasets, continue to affect the interpretation of data and the conclusions drawn in social sciences.
– While AI can spot some errors and assist in research, it still makes mistakes, carries its own biases, and lacks consistency across models without direction from human researchers, emphasizing the continued importance of human expertise.

The emphasis on human oversight and the recognition of both the strengths and weaknesses of AI in social science research reflect the nuanced view presented in the article.