Imitating, Learning, Dreaming: Machines, Conscious and Unconscious


We often get confused when we talk about what machines know about us and how they learn it.  This confusion can originate in our tendency to attribute magical powers to machines.  But processing speed and power that defies our imagination isn’t the same as consciousness or intelligence.  Projecting autonomy and agency on machines is still a function of  our overactive imagination (or perhaps an oversimplification of what autonomy and agency mean in the first place).  We get a little obsessed by what machines know without interrogating what it means to know anything.  The accuracy of their guesses can be uncanny, but it’s unclear whether to conclude that the machine is clairvoyant or that we’re just really, really predictable. 

When a machine autocompletes our queries, suggests new places to eat, autofills our tax forms, or navigates us around traffic jams, it’s functioning mimetically–replicating what a human might do, if the human had the data and processing speed of the machine.  We could all navigate around traffic jams, do our own taxes, or choose our own restaurants, given enough information.  In fact, it’s hard to say that navigating our own way around the traffic jam represents more agency and autonomy than having Waze or Google maps do it for us.  Either way, we’re both slaves to the rhythms of daily commutes and traffic patterns and free to make our own decisions (alter our daily schedule to miss rush hour; ride a bike or a bus; etc.).  

It can be surprising how closely machines can approximate the human, but the machine’s outputs are just its processing of our own inputs for its purposes.  Yelp’s algorithms aim to get its user to that new restaurant; Lyft aims to get its user from point A to point B in an on-demand car.  If we respond to a given stimulus, it’s not because the machine made us; it’s because the machine has observed us.  In this sense, the algorithms that determine an app’s outputs are operating instructions: they instruct the machine how to operate us. 

If technologies are truly user-centric, they’re likely to be engineered to accommodate the foibles of their users. We’ve seen, for example, how algorithms can have racist, sexist and otherwise irrational outputs that mirror that racism, sexism and irrationality of the data they were trained on (i.e. our data).  In that sense, technology is as unconscious of its flawed data-processing as we are.  Far from colonizing human cognition, technology has been colonized by human cognition (and it’s done some pretty dumb things as a result).

If technology can do our chores, it can also act out desires and neuroses for us as well. If we give our data to machines so they can learn and imitate and deliver back services, perhaps we also give them our desires, our irrationality and our baseness as well.  If it seems like our apps know us better than we know ourselves, perhaps it’s just reminding us of psychoanalysis’s founding insight: we don’t know ourselves very well at all. 

Philip K. Dick asked if Androids Dream of Electric Sheep, but recently I’ve been wondering if we’re the sheep that the machines dream of.  We populate the machine unconscious with the mess that is us—our highest callings and basest lusts; our tribal hatreds and our universalizing dreams; our conspiratorial nightmares and our technicolor hallucinations.  In other words, the machine is just the latest instantiation of humanity: its defenses, anxieties, desires, longing, etc.

The machine both passively receives our humanity and offers us the products of our acting out—Google doesn’t judge when we search for ourselves (or our ex-), it just reflects us back to ourselves; Facebook doesn’t care if it brings out the best or the worst in us; and Pornhub and Upworthy equally need our clicks to survive.  We feed the machine, and it processes our bits and behaviors into value for the survival of its platform.  When it stitches together the Gesamptkunstwerk of our era, will it form a world captured in its entirety by our jpegs and Snapchats?  While we’re busy worrying about implanted technologies and GMO foods, will CRISPR become capable of shuffling our genome as if it were no more than a deck of cards? And while we fret about the singularity, has algorithmic high-frequency trading already become a rogue AI in its own right (O’Reilly)?

In the wake of a cyborg market gone awry, and a cyber war that destabilized our democracy, traditional resistance is futile: Occupy Wall Street and the Women’s March, (though moving touchstones of humanity in the face inhumanity) are fighting yesterday’s battle.  Resistance is futile until we recognize that we’re our own worst enemy.  In the words of an old chestnut (itself equal to the psychoanalytic dialectic of self and other), we have met the enemy, and he is us.

Graphic: Transit Film ; Friedrich Wilhelm Murnau Stiftung ; an UFA production ; distributed by Parufamet ; director, Fritz Lang ; [produced by Erich Pommer] ; screenplay, Thea von Harbou. Metropolis. New York, NY :Kino International, 1927.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s