Michael Varnum, an associate professor at Arizona State University, explores the groundbreaking use of AI to simulate ancient minds for social psychology studies. They discuss how feeding historical texts to chatbots can unlock insights about past cultures. Varnum highlights the impressive parallels between AI-generated data and real human responses, while also addressing potential biases in training datasets. The conversation dives into how diverse cultural perspectives can enhance AI's understanding of historical psychology, shedding light on underrepresented groups and their contributions.
AI tools, particularly language models, can simulate ancient cultural perspectives, offering new insights into historical psychology and societal behaviors.
The discussion highlights the need to address inherent biases in historical data through complementary methods, ensuring a more balanced understanding of past civilizations.
Deep dives
Exploring the Use of AI in Understanding Historical Psychology
The potential application of artificial intelligence in exploring historical psychology is a key focus. Current limitations in understanding the behaviors and values of past civilizations are addressed, as traditional methods rely on indirect proxies, such as archival data. AI tools, particularly language models, offer a new avenue to simulate data from ancient cultures, allowing for insights into the mentalities of groups like Romans or Vikings. This opens up exciting possibilities for understanding cultural change by providing data reflective of ancient populations' perspectives.
Challenges of Historical Bias in Data
The challenge of inherent bias in historical data is critically examined, particularly given that most records reflect elite, literate perspectives from the past. The discussion highlights that writing is a recent development in human history and that only educated individuals contributed to surviving texts. To mitigate these biases in AI models trained on historical data, researchers consider complementary methods, including fine-tuning models to account for socioeconomic factors. Despite these challenges, the conversation emphasizes the importance of combining multiple approaches to achieve a more balanced understanding of past societies.
AI's Future Role in Psychological Research and Hypothesis Generation
The future potential of AI in psychological research includes generating new hypotheses and insights, expanding the boundaries of traditional methodologies. Encouragingly, recent findings show that AI can produce compelling ideas for social psychology research, which are perceived as valuable by human researchers. This suggests a transformative role for AI, not only in simulating participants and analyzing data but also in contributing creatively to the research process. The integration of AI is viewed as a tool that can complement human researchers' work, enhancing the exploration of both ancient and contemporary psychological phenomena.
Writings and records are how we understand long-gone civilizations without being able to interact with ancient peoples. A recent opinion paper suggested we could feed chatbots writings from the past to simulate ancient participants for social psychology studies. Similar survey experiments with modern participant data closely matched the outcomes of the real people they were based on. We speak with the opinion paper’s co-author Michael Varnum, an associate professor at Arizona State University, about what the limits of this spooky proposal are and what the ghosts of cultures past could teach us today.
Recommended reading:
“Large Language Models Based on Historical Text Could Offer Informative Tools for Behavioral Science,” by Michael E. W. Varnum et al., in Proceedings of the National Academy of Sciences USA, Vol. 121, No. 42, Article No. e2407639121; October 9, 2024
E-mail us at sciencequickly@sciam.com if you have any questions, comments or ideas for stories we should cover!
Discover something new every day: subscribe to Scientific American and sign up for Today in Science, our daily newsletter.
Science Quickly is produced by Rachel Feltman, Fonda Mwangi, Kelso Harper, Madison Goldberg and Jeff DelViscio. This episode was hosted by Rachel Feltman. Our show is edited by Jeff DelViscio with fact-checking by Shayna Posses and Aaron Shattuck.