BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:dc590c0f117e240c76925060b955f058
DTSTAMP:20260626T094847Z
SUMMARY:Dean’s Research Colloquium
DESCRIPTION: \n\nThe Dean’s Research Colloquium gathers faculty members f
 rom all\ndisciplines within the LeBow College of Business. The primary goa
 l of\nthe colloquium is to foster interdisciplinary discussions on subject
 s\nthat have significant and far-reaching consequences for research in\nec
 onomics and business.\n\nOur inaugural colloquium is on “Using Generativ
 e AI for Research”\nand will feature AYELET ISRAELI\, PHD\, from Harvard
  Business School.\nProfessor Israeli will present her innovative paper whi
 ch shows how\nlarge language models (LLMs) like ChatGPT can be used to gen
 erate\nsynthetic consumer responses to typical market research questions.\
 nFollowing the presentation\, a panel discussion among LeBow faculty\nwill
  explore the diverse applications of LLMs for research in\neconomics and b
 usiness\, ranging from pre-testing survey instruments to\ncopy editing art
 icles.\n\nFEATURED SPEAKER:\n\nAyelet Israeli\, PhD\n[https://www.hbs.edu/
 faculty/Pages/profile.aspx?facId=766753]\nMarvin Bower Associate Professor
 \nHarvard Business School\n\nPAPER:\n\nBrand\, Israeli and Ngwe (2023WP) U
 sing GPT for Market Research Large\nlanguage models (LLMs) have quickly be
 come popular as labor-augmenting\ntools for programming\, writing\, and ma
 ny other processes that benefit\nfrom quick text generation. In this paper
  we explore the uses and\nbenefits of LLMs for researchers and practitione
 rs who aim to\nunderstand consumer preferences. We focus on the distributi
 onal nature\nof LLM responses\, and query the Generative Pre-trained Trans
 former 3.5\n(GPT-3.5) model to generate hundreds of survey responses to ea
 ch\nprompt. We offer two sets of results to illustrate our approach and\na
 ssess it. First\, we show that GPT-3.5\, a widely-used LLM\, responds to\n
 sets of survey questions in ways that are consistent with economic\ntheory
  and well-documented patterns of consumer behavior\, including\ndownward-s
 loping demand curves and state dependence. Second\, we show\nthat estimate
 s of willingness-to-pay for products and features\ngenerated by GPT-3.5 ar
 e of realistic magnitudes and match estimates\nfrom a recent study that el
 icited preferences from human consumers. We\nalso offer preliminary guidel
 ines for how best to query information\nfrom GPT-3.5 for marketing purpose
 s and discuss potential limitations.\nRead full text\n[https://papers.ssrn
 .com/sol3/papers.cfm?abstract_id=4395751].\n 
DTSTART:20240531T173000Z
DTEND:20240531T190000Z
LOCATION:Gerri C. LeBow Hall\, 3220 Market Street\, Room 033\, Philadelphia
 \, PA 19104
END:VEVENT
END:VCALENDAR