Suggestions
Ben Edelman
Principal Consultant at Dun & Bradstreet
Ben Edelman is the Chief Economist for Web Experiences, Strategy, and Policy at Microsoft, a position he has held since May 2018. In this role, he provides practical and actionable advice on various topics at the intersection of software, strategy, economics, and public policy. His focus is on developing new products and features that are economically sound and offer sustainable competitive advantages.1
Professional Background
-
Harvard Business School: Prior to joining Microsoft, Edelman was an Associate Professor from July 2012 to June 2018, where he studied and taught the economics of online markets. He combined software engineering with legal and economic analysis to understand market functions and improvements.1
-
Legal Practice: He also owned the Law Offices of Benjamin Edelman from November 2005 to June 2018, focusing on class action litigation for small businesses and consumers. His notable work includes uncovering significant IT flaws in the airline industry that resulted in refunds for affected passengers.1
-
Consulting: Edelman founded Edelman Consulting in November 1998, providing consulting services related to internet systems and software, including security and accountability. His work led to major findings in affiliate marketing fraud that prompted criminal investigations.1
Education and Early Career
Edelman's academic journey includes research roles at the Berkman Center for Internet & Society, where he focused on internet architecture and policy issues. He has been recognized for his contributions to understanding online market dynamics and has authored numerous teaching cases at Harvard Business School.1
Overall, Ben Edelman's diverse expertise spans economics, law, and technology, making him a significant figure in shaping strategies at Microsoft.
Highlights
Microsoft's AI and Productivity research initiative today released its first public report. We seek to measure and improve the productivity impacts of LLM-powered tools, as well as understand and mitigate any externalities. Report here: https://lnkd.in/genMvcKY
I was pleased to lead three studies discussed in this report:
In https://lnkd.in/gP7zD4Mc , we provided 310 subjects with Microsoft 365 (Word, Outlook, Teams, etc.). Half the subjects got normal tools; half got AI-enhanced versions. We assigned them a range of tasks including information search, catching up from a missed meeting, and drafting prose. Those with AI were stat sig faster, a difference that holds both on its own and when controlling for accuracy/quality.
In https://lnkd.in/gK_vqVRM , we tested the prerelease Outlook feature "Sound Like Me," which trains a LLM's writing style on a user's Sent Items folder so that a user's AI-assisted messages sound like that user. Lab subjects found SLM-generated messages more clear and more concise. Comparing a message to its SLM rewrite, subjects favored the SLM rewrite. Comparing human and SLM messages, subjects struggle to tell the difference, in one specification doing worse than random.
In https://lnkd.in/gAYuirui , we asked security rookies (normal users with IT competence but no security expertise) to investigate security incidents. Half our lab subjects got the standard M365 Defender console, while half got the same tool with AI enhancements. Subjects had a range of tasks: summarizing incidents, analyzing scripts, recommending remediations. Subjects with AI tools achieved stat sig greater accuracy, holding speed constant.
Work on these studies has been a highlight of my time at Microsoft -- distinctive work both in testing these early tools, and in envisioning ways to make them even better.
