The next leap in chatbots will not come from better AI models alone. It will come from recognizing that with AI, the customer experience is no longer an engineering challenge. It’s now one of human psychology.

— Stefano Puntoni, Wharton Professor of Marketing and Co-Director of Wharton Human-AI Research, Fixing Chatbots Requires Psychology, Not Technology (Harvard Business Review)

Despite major advances in AI, customers still walk away frustrated from chatbot interactions. New research shows the problem isn’t about making chatbots smarter — it’s about making them more psychologically aware. By using simple behavioral insights, companies can turn frustrating interactions into customer loyalty drivers.

Read more via HBR → (subscription may be required)

Spotlight Forum on Behavioral Insights for AI Policy & Regulation

Wharton Human–AI Research’s mission is to support rigorous, interdisciplinary research at the intersection of intelligent technology, business strategy, and human behavior. As part of this mission, we were thrilled to sponsor the latest volume of Behavioral Science & Policy. This publication features a forum on behavioral insights for AI policy and regulation, showcasing research on robo-advising, AI-assisted policing, algorithmic audits, and more.

Read the full issue at Behavioral Science & Policy.

“It’s like forcing yourself to take the stairs instead of taking the elevator.”

— Hamsa Bastani, associate professor of Operations, Information and Decisions via The Wall Street Journal

In a personal reflection for The Wall Street Journal, Wharton professor Hamsa Bastani explores how overreliance on generative AI may dull our cognitive edge—and why she now intentionally chooses friction over ease to stay sharp and engaged.

Read more via WSJ → (subscription may be required)

One might view learning through LLMs rather than web search as analogous to being shown the solution to a math problem rather than trying to solve it oneself.

— Shiri Melumad, associate professor of Marketing, “Experimental Evidence of the Effects of Large Language Models versus Web Search on Depth of Learning”

In a series of experiments with over 4,500 participants, Professor Melumad finds that using large language models like ChatGPT can lead to shallower learning, reduced originality, and lower-quality advice compared to traditional web search. The research suggests that AI tools may make knowledge acquisition more passive—potentially diminishing how deeply we engage with and retain information.

Read the full working paper →

“GenAI offers firms unprecedented opportunities to understand and engage with customers, better assess the competitive environment, and extend data-driven decision-making.”

— Stefano Puntoni, Faculty Co-Director of Wharton Human-AI Research, via Harvard Business Review

In this new HBR feature, Professor Puntoni and his co-authors present a clear framework for how generative AI is revolutionizing market research—leveraging synthetic data, digital twins, and always-on insight engines to unlock deeper customer understanding.

Read the full piece → (subscription may be required)