Arnav Lal collects water at Playa Baquerizo on San Cristóbal Island as a sea lion watches.
(Image: Lisa Mattei)
2 min. read
By testing six large language models on graduate-level questions in science, engineering, and law, researchers from the Wharton School’s Generative AI Labs (GAIL) find that so-called “expert personas” deliver no consistent boost in accuracy, and in some cases worsen the results. The pattern holds across most models, suggesting organizations may get more value from how they use AI than from how they prompt it.
This cuts against widely-promoted guidance from leading chatbot makers—including OpenAI, Google, and Anthropic—encouraging users to assign roles, with prompts that cast the model as a “math teacher” or “tax expert.” But the evidence suggests otherwise.
“When LLMs first came out people assumed personas would really help, but it matters less today,” says Lennart Meincke, a research fellow at Wharton’s Mack Institute for Innovation Management who co-authored the paper.
The study is the fourth in Wharton’s “Prompting Science” series examining how different prompting techniques shape AI performance. It is co-authored by Savir Basil, Ina Shapiro, GAIL senior fellows Dan Shapiro and Lilach Mollick, and management professor Ethan Mollick.
The findings point to a different way of working with AI, coming after a period in which “prompt engineering” was widely seen as a way to get better results, even spawning a new role focused on crafting the right instructions.
Instead, the paper suggests organizations are likely to get more value from how tasks are framed for AI, what gets fed in, and how results are checked, rather than from layering personas on top of prompts.
Read more at Knowledge at Wharton.
From Knowledge at Wharton
Arnav Lal collects water at Playa Baquerizo on San Cristóbal Island as a sea lion watches.
(Image: Lisa Mattei)
nocred
nocred
Nhlanhla Mavuso of Fluid Silicon at work in the Moore Building.
nocred