Hi! PARIS Workshop on AI and Digital Economy
Participate
Friday, 21 April 2023 on HEC Campus, Building S228
The first Hi! PARIS Workshop on AI and Digital Economy will take place on Friday 21 April, 2023 from 8:45 am in S228. Our five guest speakers and the program are listed here.
Event is free for all members affiliated with Hi! PARIS and PhD students.
The registration fee below applies to external participants.
-Conference fee 30 euros also includes light breakfast, lunch and afternoon coffee and networking opportunities.
-Conference fee 90 euros also includes light breakfast, lunch and afternoon coffee, cocktail and dinner at HEC Paris Le Château and networking opportunities
Dear all,
The first Hi! PARIS Workshop on AI and Digital Economy will take place on Friday 21 April, 2023 from 8:45 am. Our guest speakers and the program are listed as below:
8:45 - 9:35 Welcome Coffee & Registration
9:35 - 9:45 Opening Words: Andrea Masini – HEC Dean of Faculty & Research
9:45 - 10:45 Key Note 1: Anindya Ghose from NYU Stern
Title: Building An AI-Based Organization: A Multi-Pillar Data-Driven Framework
Abstract: This talk will present the four critical pillars of data driven decision making in the AI era, the five essential elements of succeeding with data, how can companies win with AI in a world where many are selling snake oil, what kinds of data are relevant and critical, and what are the trade-offs faced by companies in the AI era. This would draw upon my 20 years of experience consulting for and research collaboration with more than 45 companies in 5 continents on their digital transformational journeys, and 10 years of leading the data science masters program at New York University.
10:45- 11:00: Coffee Break
11:00 - 11:30 Company Presentation: Alex Rykhva from Louis Dreyfus Company - Head of Data Science and Analytics
Title: “Data Science in Commodity Trading”
11:30 - 12:15 Research Presentation: Prasanna 'Sonny' Tambe from the Wharton School
Title: Deep Capital for Deep Learning
Abstract: We provide theory and evidence connecting successive vintages of IT investment to the marketplace advantages enjoyed by large, digitally intensive firms. Laggard firms incur additive adjustment costs when catching-up with frontier firms. Since firms use other technologies to produce frontier technological intermediate goods, advantages in earlier vintages of technology capital "stack" in terms of producing frontier intermediates. We take this model to a new panel data set on firms' technology investments. Using the launch of 80 different technology skills as natural experiments, we estimate the effect of technology improvements on AI investment. We demonstrate a progression in stages of major technology investments over the last two decades -- databases, networks, cloud computing, data science, and AI -- and demonstrate that adjustment costs across stages contribute to an effect in which firms with more developed digital capabilities face lower costs to install AI technologies.
12:15 - 14:00 Lunch at Best Western
14:00 - 15:00 Key Note 2: Feng Zhu from Harvard Business School
Title: Why Some Platforms Thrive and Others Don't
Abstract: In the digital economy, scale is no guarantee of continued success. After all, the same factors that help an online platform expand quickly—such as the low cost of adding new customers—work for challengers too. What, then, allows some platforms to fight off rivals and grow profits more successfully than others? In this presentation, I will discuss several research projects aimed at understanding what drives platform success. The findings also have important implications for policymakers looking to regulate the platform sector effectively.
15:00 - 15:15 Coffee Break
15:15 - 15:45 Company Presentation: Maxime Patte from Veesual - CEO & Co-founder
Title: "From an impressive technology to a product people want to buy: the real challenge of Generative AI in fashion"
15:45 - 16:30 Research Presentation: Ananya Sen from CMU
Title: The Value of External Data for Digital Platforms: Evidence from a Field Experiment on Search Suggestions
Abstract: Firms increasingly leverage external data to unlock improvements in products and services, but measuring the value of external data is challenging. Collaborating with a large Chinese technology company, we analyze a randomized field experiment where we manipulated access to the market leader’s application programming interface (API) to measure the causal impact of depersonalized external data on click-through rate (CTR) for the focal company's nascent search product. We report three main findings: First, compared to the baseline with access to the market leader’s API, API removal leads to a 4.6\% decrease in CTR on average for search suggestions. Second, the negative effect due to API removal is more prevalent among heavy users and for both mainstream and niche content. Third, the magnitude of this negative effect, in the long run, is half as much as what we would have obtained with a short-term experiment. This research informs managers of whether and how the market leader’s data affects a smaller player's product performance. It further sheds light on policies such as the Digital Markets Act that proposes privacy-preserving data sharing by large digital platforms and a recent debate on whether big data undermines market competition.
16:45 - 17:30 Research Presentation: Anuj Kumar from U. of Florida
Title: Upgrading Education with AI-enabled Knowledge Diffusion
Many EdTech solutions aim to improve learning by offering educational content aligned with students’ learning needs. However, it benefits those students who are motivated and have adequate support to learn from the personalized content. Since students from high-income families have higher access, motivation, and help to learn from such software, they benefit more than their underprivileged counterparts.
I designed an EdTech platform (Epic app) that offers students personalized content, motivates them to consume it, and provides peer support to learn from it. The Epic app organizes students with varying abilities into balanced teams and conducts educational tournaments between them in classrooms. The app’s AI engine identifies students’ knowledge gaps based on their homework performance and intelligently connects them to team members who have done well on those topics. Winning the tournament as a team motivates students to learn concepts taught in the class and incentivizes them to help team members in areas of their knowledge gaps. Such peer-led knowledge diffusion should result in higher and more uniform learning.
I conduct a large-scale experiment on 2000 students in 20 Indian schools to examine the effect of peer-led knowledge diffusion. I randomly divide students into three groups. While students in the first and second groups access the app in teams of four and one, those in the third group do not use the app. Differences in cognitive achievements between the first and third group students would reveal the effect of peer-led knowledge diffusion on overall learning and achievement gaps among students. Comparing the achievement distributions for students in the three groups would further unmask whether (1) high-ability students suffer from sharing knowledge with their low-ability counterparts and (2) low-ability students are worse off by participating in the tournament individually.
17:30 - 19:00 Cocktail at HEC Le Château (optional)
19:00 - 21:00 Dinner at HEC Le Château (optional)