It is one thing to continue smoking despite general statistics that connect smoking with lung cancer. It is a very different thing to continue smoking despite a concrete warning from a biometric sensor that has just detected seventeen cancerous cells in your upper left lung. And if you are willing to defy the sensor, what will you do when the sensor forwards the warning to your insurance agency, your manager, and your mother?
Who will have the time and energy to deal with all these illnesses? In all likelihood, we could just instruct our health algorithm to deal with most of these problems as it sees fit. At most, it will send periodic updates to our smartphones, telling us that ‘seventeen cancerous cells were detected and destroyed’. Hypochondriacs might dutifully read these updates, but most of us will ignore them just as we ignore those annoying anti-virus notices on our computers.
The drama of decision-making
What is already beginning to happen in medicine is likely to occur in more and more fields. The key invention is the biometric sensor, which people can wear on or inside their bodies, and which converts biological processes into electronic information that computers can store and analyse. Given enough biometric data and enough computing power, external data-processing systems can hack all your desires, decisions and opinions. They can know exactly who you are.
Most people don’t know themselves very well. When I was twenty-one, I finally realised that I was gay, after several years of living in denial. That’s hardly exceptional. Many gay men spend their entire teenage years unsure about their sexuality. Now imagine the situation in 2050, when an algorithm can tell any teenager exactly where he is on the gay/straight spectrum (and even how malleable that position is). Perhaps the algorithm shows you pictures or videos of attractive men and women, tracks your eye movements, blood pressure and brain activity, and within five minutes ejects a number on the Kinsey scale.6 It could have saved me years of frustration. Perhaps you personally wouldn’t want to take such a test, but then maybe you find yourself with a group of friends at Michelle’s boring birthday party, and somebody suggests you all take turns checking yourself on this cool new algorithm (with everybody standing around to watch the results – and comment on them). Would you just walk away?
Even if you do, and even if you keep hiding from yourself and your classmates, you won’t be able to hide from Amazon, Alibaba or the secret police. As you surf the Web, watch YouTube or read your social media feed, the algorithms will discreetly monitor you, analyse you, and tell Coca-Cola that if it wants to sell you some fizzy drink, it had better use the advertisement with the shirtless guy rather than the shirtless girl. You won’t even know. But they will know, and such information will be worth billions.
Then again, maybe it will all be out in the open, and people will gladly share their information in order to get better recommendations – and eventually in order to get the algorithm to make decisions for them. It starts with simple things, like deciding which movie to watch. As you sit down with a group of friends to spend a cozy evening in front of the TV, you first have to choose what to see. Fifty years ago you had no choice, but today – with the rise of view-on-demand services – there are thousands of titles available. Reaching an agreement can be quite difficult, because while you personally like science-fiction thrillers, Jack prefers romantic comedies, and Jill votes for artsy French films. You may well end up compromising on some mediocre B-movie that disappoints all of you.
An algorithm might help. You can tell it which previous movies each of you really liked, and based on its massive statistical database, the algorithm can then find the perfect match for the group. Unfortunately, such a crude algorithm is easily misled, particularly because self-reporting is a notoriously unreliable gauge for people’s true preferences. It often happens that we hear lots of people praise some movie as a masterpiece, feel compelled to watch it, and even though we fall asleep midway through, we don’t want to look like philistines, so we tell everyone it was an amazing experience.7
Such problems, however, can be solved if we just allow the algorithm to collect real-time data on us as we actually watch movies, instead of relying on our own dubious self-reports. For starters, the algorithm can monitor which movies we completed, and which we stopped watching halfway through. Even if we tell the whole world that Gone With the Wind is the best movie ever made, the algorithm will know we never made it past the first half-hour, and we never really saw Atlanta burning.