All Your Choices Belong to Us
What does the fast approaching world of the “Internet of things”(IoT) and Algorithmic based automated control mean for decision making and free will?
Technology for many years has been about opening the door to new choices and options for nearly everything in our daily life. From travel planning to social interaction to shopping to photography, the constant expansion of the universe of information enabled by technology and specifically the Internet has been incredible. It has also become phenomenally overwhelming. The explosion of posts, tweets, and pins combined with the ubiquity of smartphones has begot another recent trend of artificial intelligence to reduce volume and narrow the selection to only the things which are deemed ‘most important’ to us, as best that can be determined through the use of algorithms and smart filters.
In many cases these algorithmically derived filters are essential. You couldn’t possibly view all the posts for every friend in your Facebook feed. You can’t read all the news from around the world on any single topic. In the most basic way, you are giving up some control for a promise of increasingly relevant content. The tradeoff is clearly positive.
But what happens however when this basic principle is applied not to news or content but to more critical areas of our daily life? Areas that might truly have life or death consequences? How comfortable are we in moving from a world where computers make recommendations to one where computers passively automate choices to, eventually, where they actively restrain or prevent us from making our decisions and remove control entirely?
Several recent conversations have left me wondering if the dawn of compute and connected devices, aka the Internet of Things, have pushed us to a world where free will and decision making is beginning to fade into the distance.
The first conversation was with my current boss, who recently attended Kara Swisher’s Code Conference in Palm Beach where Google revealed the latest ‘self-driving’ car to the world. While many mocked the comical appearance (my favorite was that it looked like the buggies used to bring baseball pitchers in from the bullpen), the most striking aspect for me was the removal of all standard controls except for a giant (panic?) button. Steering wheel? Nope. Gas, er, I mean accelerator pedal? Nope. “Software and sensors do all the work”.
This layout is in stark contrast to the first and second generation Google self-driving cars, a Prius and Lexus SUV respectively. Those cars were essentially standard vehicles with massive amounts of computational and automated driving features added on. Notably at any time the driver could take over control from the network and revert to a more traditional operation. The new car…not so much.
The evolution was however fairly logical. We went from a world of recommendation (GPS systems) to automated recommendation (driverless capabilities on a standard vehicle) to active restriction, no vehicle controls available to the driver whatsoever. Essentially you are completely at the mercy of the Internet at that point. Wow.
Another thing that struck me was that many people, myself included, love driving. Its not simply a chore or a means to an end. It is in fact an experience in-and-of-itself. And, depending on your choice of vehicle, a very rewarding one at that.
I can’t even stand driving an automatic transmission, never mind an auto-everything. The freedom of the roads, the option to go anywhere at anytime, is something intrinsic tied to the mindset of several generations. It is essentially free will. And Google, among others, are now developing technology to take that away, though admittedly not as a primary or even secondary objective.
The second discussion, or more accurately presentation, that pushed me further to consider the societal impact of automated tech was a great presentation about the ‘Internet of Things’ given by my old boss and current Sequoia partner Omar Hamoui. Omar’s opinion of why smartphones have achieved such rapid adoption and what this means for other technology areas was simple and stunningly accurate, as usual.
The phone, in its basic form factor, was known to nearly everyone. By adding compute and connectivity capabilities, the device simply become smarter and more ingrained in people’s lives. Following this line of thought a similar technology adoption curve will likely spread to many other objects in our daily lives; thermostats, light bulbs, lawn watering equipment, BBQ propane tanks (OK, that is just my own personal request for the people that make things on the Internet of Things). By adding compute and connectivity to everything, manual operated devices become intelligent, and thus, hopefully more convenient and valuable.
Viewed in this way, the self-driving car is an inevitable outcome; adding compute and connectivity into a device that we already know. What I think is critically different, at least at the moment, is that there isn’t a substantial new set of capabilities that I am being offered. I know how to drive. In fact I believe, as do most people, that I’m a fairly exceptional driver. Claims of reduced traffic and transit times seem spurious and likely unobtainable in the near term at least; the only way these will be achieved is if a large percent (majority?) of drivers are also using driverless technology. So what am I getting for my lose of control?
I started to consider the further evolution of this trend and in what other areas of daily life the same compute and connect capability may lead to further restriction or limitation of control. For instance, in a world where thermostats are completely automated for everyone, and the data about these settings are centrally, what happens if I decide to set my home temperature 10, 15, 20 degrees above the ‘recommended’ norm? Will i be allowed to do this? Will I be subject to ‘surge’ pricing?
Clearly ‘home automation’ has lots of benefits. The global aggregate of energy savings from ‘smart’ light bulbs, heating systems and lawn watering is probably astronomical. I’m sure someone has measured this already. I’m more focused, as I think others should be, on the unintended downstream impact for control and inevitable system failure. After all, we are are talking about technology. There will be bugs. And at what cost?
Moving even further along the life impacting technology spectrum is the emerging health products. Uploading my activity to an app or website will seen be eclipsed by near constant bio-metric monitoring. What happens when my iHealthBand not only monitors my glucose level but then reports immediately to my physician (or when I eat a donut? Or I ‘accidentally’ send my blood alcohol content over .08? Will I be prevented from doing so in some way? Perhaps deactivating my the payment feature in my iBand for certain merchants? Or more specifically when certain products are purchased? What crosses the line of incentivized behavior vs. preventive monitoring?
Engineers are focused, as they should, on building the future. Writers and journalist often seem to write about the impact of technology once products are released into the world. In the case of the Internet of Things and algorithmically based life tools, some forethought and pre-emptive boundaries seem at least worth discussion. Do we need regulation? The Federal Software Administration perhaps? Don’t laugh the FDA had very little oversight until the mid 20th century after a series of poorly tested drugs were released with horrible unintended side effects. Maybe we need a layer of independent testing and product validation companies?
My only conclusion is that we will in fact lose control of many basic decision making opportunities and as a result also lose our desire to make decisions. We will simply ‘expect’ decisions to be made for us. I guess the real question is if the technology under development will be designed flawlessly enough so that I forget I ever wanted to make those decisions in the first place…
