I recently found myself in a conversation with some friends (thanks again to Dan M. (@OhMDee), @zcichy, and Eric (@mobile_reach) I made on Twitter (yes, you can make friends on Twitter). Our conversation was a frequently productive and sometimes frustrating back and forth on Apple’s privacy position and what risks it may have on their future competitiveness with services, namely AI/Siri.
While Apple is not going to be a pure play services company, there is no doubt services will play a much larger role in consumer experience in the coming years. It is reasonable to believe one’s ability to compete in features around machine learning and, eventually AI, will depend on the depth and quality of data acquired to train your networks and AI assistants. So let’s start by looking at Apple’s relationship with customer behavior data.
Is Apple Getting Enough Useful Data?
Apple’s relationship with customer data has always been clear. If you agree to share analytics/diagnostic information with Apple, you are opting-in to share some data with Apple. They are upfront about what this data is used for as they state very clearly they are collecting data on user behavior which will be used to help make products and services better in the future. Pointedly, a key difference here, as opposed to many other services, is even if you opt out of sharing data, you still get to use the full features of the service. With their advancement in tactics around privacy, including differential privacy, they are purposefully anonymizing that data so any information collected — things you say to Siri, what apps you install, what news you read in Apple News, etc. — can never be tied back to the individual. The technical term for this is Personally Identifiable Information. Apple’s goal is to make it so no information collected can ever be tied to personally identifiable information. While no one will dispute Apple’s attempts to go above and beyond to protect our privacy is admirable, there are a few concerning points I’d like to call out.
First, we have acknowledged Apple is using information about us to make their products and services better. But we simply have no idea how much information is being collected or analyzed. The rub is Apple’s services are progressing (or, at least, feel like it quite often) at a rate much slower than other companies who do collect and analyze more customer behavior data like Google, Facebook, and Amazon. There is no doubt Siri still has advantages in global language support and integration across all of Apple devices while the competition still has limitations. While Siri is certainly competitive with Google Assistant and Amazon’s Alexa (none of them are perfect yet or without faults) you have to admit both are pretty advanced and comparable to Siri in many ways. Both Google’s Assistant and Amazon’s Alexa have been on the market less than a year while Siri has been on the market in five years. Despite technical advancements in machine learning and natural language processing during that four year gap that benefited Amazon and Google, there is no doubt in my mind their massive data sets on behavior was useful in feeding their backend engine to reach near parity with Siri from a machine intelligence standpoint.
Look at my brief time on Android using Google Now compared to Apple’s Proactive and now Siri apps. Both are supposed to be learning about me and making intelligent and contextual recommendations that sometimes work but more often than not, don’t. I’ve been on iOS since 2007 yet, a few months on Android yielded better contextual and relevant recommendations on a more consistent basis than both Proactive and Siri. This observation leads me to believe competing services are learning and getting better, faster by using more of my behavior data to analyze than Apple. The only thing I can think of is it’s because of Apple’s desire to have a hands off approach to my data.
All of the above points lead me to my final observation. I believe it is essential that Apple is competitive with services like Siri, and many others, against those whose business models depend on more on data collection than Apple’s. While I don’t believe Google and Facebook are the bad actors Apple portrays them as (and neither do consumers via evidence from our surveys), the bottom line is their business model, the financial lifeblood of their company, depends on their ability to sell advertising with the data they collect on customers using their service. Where Apple’s business model does not depend on using customer data collection to sell advertising, it is necessary for their model to make products and services that delight their customers. Within this viewpoint, Apple is already a trusted entity with our privacy since their business model does not necessitate mining that personal information. Based on some recent research we did, Apple customers overwhelmingly listed Apple as the top company they trusted with their privacy over companies like Microsoft, Google, Amazon, Samsung, Facebook, etc.
However, getting useful and good behavioral data is essential for Apple to make better products and services and, more importantly, compete with those services down the road. I’d almost prefer that, instead of Apple’s stance being not only to collect as little data as necessary and also to universally anonymize that data, they would simply say, “Trust us with your data. We will keep it safe and secure and we will deliver you superior products and services because of it.” I could also be satisfied with a hybrid approach where, for the most security conscious customers, Apple gives them the option to keep the existing privacy protocol as well as their differential privacy techniques, but also allow others to opt-in to giving them more data so that things like Siri, News, Apple Music, etc., benefit from that data and thus, deliver those customers a much more personalized and useful service. With some of the recent changes in iOS 10.3, I feel they are getting closer to exactly this scenario.
My genuine concern with Apple solely relying on an “above and beyond” approach to consumer privacy is we don’t know yet if this process will work and the existing evidence causes a great amount of speculation. My concern is they are mortgaging their future competitiveness around things like AI and better services holistically with this stance. Thus I view it as somewhat risky even if it seems like the right thing to do. If their approach does not work and their services truly not compete, some of their customers may use solutions from competitors whose business models open the door for them to be irresponsible with our data. If that happens, the customers lose because Apple — and I include Microsoft in this statement — have the least motivation to be irresponsible with our privacy. Their business models do not depend on directly monetizing that data. Say Google becomes the AI agent of the future and, all of a sudden, they fall on hard times and the only way to right the ship is to compromise or alter their privacy stance to keep making money. While it is only a hypothetical, it is still a valid concern if a free service monetized by ads becomes the majority services monopoly in the future.
I truly hope Apple is continuing to take a hard look at how their services compete in the market against comparable ones. Should there need to be some pivots on how data is collected and used to compete, I think the market would be OK with that. They are no doubt doing the right thing for their customers but, if going above and beyond with differential privacy yields non-competitive and thus less relevant services, then it will be all for nothing if their services aren’t used and can’t protect their customers.