We don’t remember much. Compared to computers, we remember almost nothing.
Computers are way better at recording and keeping visual information than human minds. I think I actually have a pretty good memory for things like names, faces and ideas; and when I walk down the street I think I’m aware of most things I see. But if you quizzed me about my journey ten minutes later, I wouldn’t be able to remember all the parking signs, the store names, or even the faces of all the people I walked by.
But a computer can.
If you provide an AI-powered computer with a lot of visual data, say all the cameras on an autonomous car, that computer could filter out different combinations of pixels that are meaningful and categorically save them just as it would a word document.
In other words, AI software can take what looks like a bunch of unmanageable, noisy data and turn it into something useful: signal.
Noise is becoming manageable information, and there is a lot of noise to play with.
Think of your morning 5 years ago. You would wake up, make breakfast, go to your car, drive to work, walk from the parking lot to the office, get into your work station, maybe chat with your colleague, and THEN start working on your computer or calling customers.
Life in a major city in the developed world looks pretty different now. People are waking up with Amazon Echo, your TV watches you as you move around your kitchen, your phone registers when you wake up and starts tracking the news and emails you’re looking through, Fitbit monitors your biometrics, a connected car or tesla records everything you drive by and all your stops, and your BlueTooth device is listening in to every conversation.
Drones, satellites, self-driving cars, they’re just the beginning in a connected city. Some technologists are using Wifi to create 3D holographic images of a room using the reverb and microwaves. Machines and sensors are EVERYWHERE, they’re getting more sophisticated, and they’re getting cheaper.
The Dilemma of Better Data
Even though you don’t see this data, someone, somewhere is structuring, saving, memorizing and analyzing it to reach business goals. There is very little privacy today.
Should we be worried?
Here’s the problem: A lot of that data collection, including the collection enabled by AI, is to fuel other AI-powered services. Anything “AI” depends on a huge amount of data to work. Terabytes of data. More data than any human could wrap their head around.
But depending on what the AI-powered program is trying to do, the data it needs can get pretty personal. It could be your location. Or every product you looked at before buying a shirt online. Or the actual words you type to your friends on a messaging app.
So is it ethical for companies to know that about its users?
As the co-founder of Element AI, I think businesses should leverage artificial intelligence to be the best in their industry. But as a citizen, do I really want to allow companies I’m not even familiar with to know so much about me?
Transparency is key
In theory, if we’re concerned that we don’t know what companies are doing with our data, the solution is easy: just ask. Legally, they have to give it to you.
But if you ask a company what data they collect about you, and what they do with it, they will probably refer you to their kilometre-long “Terms and Conditions” document. So that doesn’t solve anything.
In my opinion, too many companies are hiding behind pages and pages of documents. Companies are technically being transparent in terms of what the law requires. But they’re actually doing everything possible to hide the mechanism of their product.
Unfortunately, this is especially problematic for AI-centred companies. Often, state of the art AI developers can’t give a simple summary of how your data is used. Why? Because AI works in many different dimensions, processing information in levels and networks. All of this is hard for an ordinary person to understand.
So what’s the most honest path forward?
Generally, I expect people to tell me things that I should be aware of. I expect my business partners to tell me when they’ve had a really good idea, and my Boston Terrier to indicate when he’s hungry. That seems to be a pretty good system of communication.
Perhaps we should all expect the same courtesy from every app, website or other device that is happily collecting our information. (Tristan Harris's is worth checking out.)
I don’t think data transparency will happen on its own in the market. Unless it is an industry standard, no one will do it because they would risk killing their business by giving away their data to competitors. The playing field would need to be levelled, either through customers demanding visibility, or government creating an economic impact. Just last week, The Economist wrote that data is more valuable than oil. Perhaps we should be accounting and reporting it in the same way on the balance sheet.
What’s your perspective? Are there particular pieces of data you would never want made public? Should government be stepping in now instead of later?
Let me know what you think. I look forward to continuing the conversation below!
As I return to blogging, I’m starting to discover the categories I’d like to dive into. (Who knows, maybe they’ll even become chapters to a book.) I’ll continue looking at “What happens when noise becomes information” and get into some more detailed answers in terms of policy and technology. Earlier this week I also started writing about “Why AI is only getting exciting now.” Stay tuned for more and join the mailing list if you want them in your inbox.