Is he transparent enough?. Photo: Andrew Harrer/Bloomberg via Getty Images

The convenience of the smart home may be worth the price; that’s for each of us to decide. But to do so with open eyes, one has to understand what the price is. After all, you don’t pay a monthly fee for Alexa, or Google Home. The cost, then, is a subtle one: a slight psychological adjustment in which we are tipped a bit further into passivity and dependence.Â
The Sleep Number Bed is typical of smart home devices, as Harvard business school Professor Shoshana Zuboff describes in The Age of Surveillance Capitalism. It comes with an app, of course, which you’ll need to install to get the full benefits. Benefits for whom? Well, to know that you would need to spend some time with the 16-page privacy policy that comes with the bed. There you’ll read about third-party sharing, analytics partners, targeted advertising, and much else. Meanwhile, the User Agreement specifies that the company can share or exploit your personal information even “after you deactivate or cancel … your Sleep Number account.” You are unilaterally informed that the firm does not honor “Do Not Track” notifications. By the way, the bed also transmits the audio signals in your bedroom. (I am not making this up.)
The business rationale for the smart home is to bring the intimate patterns of life into the fold of the surveillance economy, which has a one-way mirror quality. Increasingly, every aspect of our lives — our voices, our facial expressions, our political affiliations and intellectual predilections — are laid bare as a data to be collected by companies who, for their own part, guard with military-grade secrecy the algorithms by which to use this information to determine the world that is presented to us, for example when we enter a search term, or in our news feeds. They are also in a position to determine our standing in the reputational economy. The credit rating agencies and insurance companies would like to know us more intimately; I suppose Alexa can help with that.
Allow me to offer a point of reference that comes from outside the tech debates, but can be brought to bear on them. Conservative legal scholars have long criticized a shift of power from Congress to the administrative state, which seeks to bypass legislation and rule by executive fiat, through administrative rulings. The appeal of this move is that it saves one the effort of persuading others, that is, the inconvenience of democratic politics.Â
All of the arguments that conservatives make about the administrative state apply as well to this new thing, call it algorithmic governance, that operates through artificial intelligence developed in the private sector. It too is a form of power that is not required to give an account of itself, and is therefore insulated from democratic pressures.Â
In machine learning, an array of variables are fed into deeply layered “neural nets” that simulate the binary, fire/don’t-fire synaptic connections of an animal brain. Vast amounts of data are used in a massively iterated (and, in some versions, unsupervised) training regimen. Because the strength of connections between logical nodes is highly plastic, just like neural pathways, the machine gets trained by trial and error and is able to arrive at something resembling knowledge of the world. The logic by which an AI reaches its conclusions is impossible to reconstruct even for those who built the underlying algorithms. We need to consider the significance of this in the light of our political traditions.
Join the discussion
Join like minded readers that support our journalism by becoming a paid subscriber
To join the discussion in the comments, become a paid subscriber.
Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.
Subscribe