Sramana Mitra: The Netflix example is not a right example, because Netflix is not a vendor that sells AI software to other people. I think this question is only reasonable when you take into account a vendor who sells any domain-specific AI software to a whole lot of different customers who could be competitors.
Netflix is running their own service and they collect data. They own the data. I don’t think that’s even a question that Netflix should be able to use that data to be able to power their recommendations better. I don’t think that’s up for questioning anyway. Other people may think about it.
In the case of SAP, Oracle, or any other vendor that is more narrowly-focused, the question is more complicated. What I was trying to ask you is, are there trends in particular domains where the answer is more obvious in either direction, especially where the software is being sold to a lot of competitors? What do we see and what are you seeing?
Mike Flannagan: I think there’s a lot of variation in individual companies that is not necessarily specific to industries. Whether you talk about healthcare or retail, different companies have vastly different views on the value of their data.
Sramana Mitra: My take is that it would also be very specific to the way the AI algorithm operates, right? If you’re talking about retail, if the AI algorithm has been able to identify me as a customer and wants to use that knowledge to target offers from multiple competing vendors, and if that data is being enhanced by all of their heuristics, that becomes a very questionable scenario from the perspective of the different competing vendors.
Mike Flannagan: That’s right. Once I provide my data to enrich a broader dataset and if that allows my competitors to identify customers, perhaps their offers are more compelling.
Sramana Mitra: That’s exactly my point. That’s the situation that I’m alerting our audience to. I think, in that case, the answer should be no. There should not be a cross-client collaboration or data usage.
Mike Flannagan: I appreciate your point about Netflix. I used that as an example that everybody could quickly relate to. If you took something like Spotify offers via an API and make music recommendations and if you think about it in today’s world, you would probably say the same as you would say about Netflix, “It’s just making a music recommendation, what’s the big deal?”
I read some research recently about a particular retailer. They are doing quite a lot of analytics on sales in their stores and evaluating how sales change based on what music is playing in the store during certain periods of the day. If you start thinking about that, you start thinking about something like the Spotify API, which seems benign, if a retailer can figure out how to build a recommendation engine that plays songs like the ones that cause my customer to buy more, that does become my intellectual property and a competitive advantage.
Sramana Mitra: Yes, sure. That’s a very good example.
Mike Flannagan: That’s when I think, some of those consumerized examples are things that may, initially, seem quite benign, but can have the potential to be turned into a competitive advantage if you think a little bit differently.