Sramana Mitra: In that scenario, if SAP strikes a deal with Spotify and develops that intelligence and then sells it to a variety of retailers, and the retailers help enhance that model, and then if you go back and sell that enhanced model to every single retailer, that becomes a very questionable scenario.
Mike Flannagan: I think that’s right because you start getting down to data that is very specific to that individual retailer. Back to the example of targeted advertising, that is about trying to pull you into my store versus into another store.
If you consider the example of music playing in the store increasing the propensity to buy or increasing the total dollar value of a purchase, that still starts with an expectation that I’ve come into your store. I could easily argue in the other direction that that may actually do nothing to pull customers away from you to a competitor. It might actually improve everybody’s ability to sell when their customers were coming to the store anyway. In that case, you could argue that it’s actually for the greater good.
Sramana Mitra: The problem is now you get into a situation where the same music is playing in every single retail store.
Mike Flannagan: Then it becomes a commodity.
Sramana Mitra: It becomes a commodity, yes.
Mike Flannagan: Then the question is, “How do you target advertising so that I can start pulling people into my store?” That’s what’s fascinating about the space of artificial intelligence. There are so many possibilities. We are just beginning to scratch the surface of what we can do.
Sramana Mitra: It’s very exciting. Can you actually take a couple of more use cases? The process that we are discussing is fascinating so if you have other interesting use cases where you’re pushing the envelope, that would be very interesting to delve into.
Mike Flannagan: I just went on a journey that led to something about the greater good. Let me start with something that I think is a greater good discussion. Every major corporation in America has a Board level topic, such as how do they improve on diversity and inclusion? If you look at the way that companies recruit, find, and interview candidates, not much has changed in the past 40 years.
Yes, resumes are digital now but we still have a recruiter who through all their best intentions has some inherent bias in the way they recruit. You have hiring managers who, in spite of their best intentions, have some inherent biases in how they evaluate candidates. You still have compensation practices that determine how much you offer someone. It’s typically based on previous salary and such.
We understand that given the fact that there is a fair amount of disparity of pay between men and women for doing the same job, if all I’m considering in making a compensation decision is salary history, there’s an inherent bias that would suggest that since women have historically been paid less, they would be paid less in a job offer. For all of those things, we understand that there is some inherent flaw in the system in the way that we do recruiting and hiring, but we still do it much the same way.
We have an opportunity now through using machine learning and artificial intelligence in the way that we evaluate resumes and candidates and determine salaries to remove some of that inherent bias. Obviously, unintended bias.