Thanks for the post, Louis. This is a very impactful application of AI/ML and one that I did not expect – I always thought AI would come for the real economy first, but this is a clear disruption of the digital economy. My biggest concern regarding the growth and application of Copilot in the medium to long term is one that you raised at the end of you post: data privacy and IP. It reminds me of how Professor Greenstein wanted to use Darth Vader’s AI-generated image for a case but had to remove it due to copyright constraints. Data privacy and IP issues will limit the availability of training data tremendously, and tech companies like Microsoft/GitHub will need to get creative to fill in the gap. Perhaps synthetic data is the solution?
Super interesting, thanks for the post. I was not aware of this interesting application of AI/ML. Based on my reading of the post, ASOS is limiting AI/ML applications to profile building of its customers and the Buy The Look feature. I wonder if they can take ML/AI a step further and apply it to their and their partner’s manufacturing processes to reduce cost and increase efficiency, while also applying the technology to help in the creative process of designing the latest apparel.
Thanks for the post, Anand. I am a Capital One customer and can attest to their seamless integration of tech in their financial product offering. An additional service they have which I use weekly and I am sure has some level of AI/ML integration is Capital One Shopping, a Chrome plug-in that screens every single e-commerce checkout page you navigate into in order to offer you discount codes and alternative sites with cheaper prices. It is one of my favorite Capital One products. It allows Capital One to collect massive amounts of data from customers browsing the internet, which can then be leveraged by advertisers to provide targeted discounts and convert high-likelihood shoppers.
Thanks for the post, Gigi! This is very interesting. I wonder what causes so different outputs in seemingly similar AI models with the exact same prompt. Is it a result of model training differences? Differences in raw data sources? We think AI models will help us solve many of humanity’s problems, but it seems one day we may be saturated by the different AI model choices at our disposal.
Thanks for the post, Laura. I wonder how the AI is improving upon itself. Does it know when it did a good job of creating an accurate image, and when it did not do a good job? Maybe the researchers could outsource this training by having users score the outputs from 1 to 10 based on how accurate the images are.
Thanks for the post, Jiwon. I wonder if AI can be trained to recognize these biases and the stereotypes that it reinforces. Is improving AI to be cognizant of societal biases and address them in outputs a lost cause that demands human intervention? Will data scientists and ML researchers need to “break” the automatic learning loop inherent in AI to fix this issue, or is it something we can train the AI itself to fix?
Thanks for the very interesting post, Elizabeth! I think it’s relevant to analyze Venmo in the context of its ownership by PayPal, which acquired Venmo in 2012. Could Venmo be a “loss-leader” to on-board clients into the broader PayPal ecosystem, where the company is actually able to generate a profit from them? Or is PayPal waiting for the right moment to fully integrate Venmo into is broader service offering, brand and ecosystem?
Thanks for the post, Amy! I was a Classpass customer and love the product. However, I was definitely an expensive customer for them given I would maximize the referral and sign-up discounts, go to the most expensive studios they offered (Flywheel at the time), and then just continue as a Flywheel customer (not a Classpass customer) once my free / promotional credits were used. Although I have not used Classpass in years, I wonder if there is a way for Classpass to fully own the customer experience, instead of just being an acquisition channel for studios.
Thanks for the post, Isa! I think your point on profitability is spot on. Rappi has burned through billions of investors’ money but has yet to prove profitability on a unit economics basis. Margins for 30-minute delivery are just not there for Rappi to validate a $5bn valuation. We have seen companies like Doordash, Deliveroo and Uber fall dramatically in value as investors realize the scale needed to get to unit economics breakeven. I believe Rappi will have to raise capital again soon, at a lower valuation, and an IPO in the near to medium term is definitely out of the question.
Thanks for the great content, Carlos!
What FCX accomplished by deploying data analytics in its mining operations is nothing short of extraordinary. A few questions come to mind:
1. Would FCX be able to develop and enact such an initiative without having to hire external consultants (McKinsey)? I think this question is important as we think about potential avenues of disruption for more ‘traditional’ industries, and how they may or may not be well-disposed to new ways of doing things (ie. data analytics).
2. How scalable are the AI models that FCX developed for the Bagdad mine? Can they quickly and efficiently be implemented in other FCX mines across the globe, or do they need months worth of work before being deployed somewhere else?
Thank you for the cool post, Paulina!
I had the opportunity to meet Kavak’s founder, Carlos, at an HBS class just this week. One of the most impressive things that he mentioned is that a majority of the customers to whom they lend money to buy used cars have never been able to request a loan through traditional financial institutions. In that sense, Kavak is having a tremendous social impact because it allows this unbanked individuals to start building their wealth by giving them their first large-scale asset: a car. Moreover, Carlos mentioned that on average the income of the people who buy their first car through Kavak doubles a year after the purchase. This means that these customers are using the car to supplement their income streams (perhaps by reducing commuting time, or driving for a ride-hailing service). This is a very clear example of how the effective implementation of data analytics can have tremendous social implications and lead to economic progress at the base of the pyramid.
Thanks for the post, Elizabeth! I was not aware that Ibotta existed. As I read about its value prop, one of the key problems I think Ibotta is uniquely positioned to solve is the accurate attribution of marketing spend to offline purchases. Today, brands target customers via digital marketing but unless customers purchase online, it is very hard to measure the ROI of that spend. With Ibotta and its Big Data engine, brands could be able to more accurately assess the actual sales generated by these ads. This potentially requires Ibotta to be connected to a user’s Google, Facebook, or Twitter accounts, but could lead to better ROI tracking of digital marketing spend.