In Netflix’s fascinating documentary, Coded Bias, we learn about racist algorithms (you know, the algorithms that can’t identify Black people’s faces as faces, the algorithms that recommend that Black men get sent to jail for longer sentences than White men because of all the years of biased data (see ProPublica), and the algorithms that sell poor people advertisements for casinos and payday lenders), but poetry was the last thing I expected to encounter. When it comes to algorithms vs. poetry, we pretty much already know which one is going to win, and it’s not going to be poetry. When it comes to social algorithms, intensity of feeling feeds the signal. And the more intense and bad the feeling, the stronger the signal. If feeling feeds the signal, poetry’s signal is too subtle to be heard, while angry click-bait will get happily get picked up every time. Joy Buolamwini, the heroine of Coded Bias, is no ordinary artificial intelligence researcher, and Coded Bias is not an ordinary film about tech dystopia. One of the most important moments in the film was when Buolamwini stopped in the middle of an advocacy meeting (where she fights to protect people from invasion of technology into their private spaces) to read a poem she had written. It was a poignant moment because poetry is probably one of the lowest-tech forms of expression out there. And, here it was, still serving a purpose in the fight against bias. Buolamwini is an activist helping communities fight coded bias. She calls herself a “poet of code” on her website.
These days, it’s all too easy to villainize the big tech giants, but much more difficult to put together a convincing argument about how we should respond to their many “disruptions.” I’ve written on this blog about Jaron Lanier’s Ten Arguments for Deleting Your Social Media Account Right Now, but Lanier still has failed to get me to quit social media, despite the fact that I agree with him entirely. Algorithmic injustice, that is to say, problems that arise when algorithms reflect our culture’s biases, are more easily addressed from the top down, than from the bottom up (that is, one person out of a billion choosing to delete a single account). In fact, Coded Bias is the rare social and cultural documentary that offers a fairly straightforward and practical solution to the problem. Buolamwini argues that there should be an “FDA for code and algorithms.” Basically, the researchers argue that we need government oversight of the algorithms being used to make decisions about our lives.
Algorithms affect virtually every aspect of our lives. Algorithms influence our search results. They influence the advertisements we are shown on social media and on the web. They influence which friends’ profiles we see first in our feeds, and which profiles we see in our feeds period. They influence whether we get credit, or don’t get credit. They influence the price we pay for insurance. They influence the criminal justice system, in that they offer sentencing guidelines and probation guidelines in some jurisdictions. They can influence hiring decisions. Algorithms can influence which college applications or resumes don’t get read, and which ones get rejected outright. They can influence who gets a seat at the table. They influence search results which shape our economic decisions and our understanding of the world.
The artificial intelligence researchers and algorithm critics in Coded Bias argue that we need an “FDA for algorithms” to review any algorithm for potential bias before it is used in public. This is just basic common sense.
Algorithms are the opposite of poetry, but in some ways they are also similar. You put raw data into an algorithm. Something mysterious happens in the “black box” of code, and data comes out. The person creating the code doesn’t know why the results came out the way they did, just that they did. The same is a little true for poetry. The poet experiences the world. Something mysterious happens. A poem is written. When the poem is good, the poet often doesn’t know how it came out the way it did.
I first encountered coded bias when I read about it in a ProPublica article years ago. I was working for criminal defense lawyers, writing SEO copy for their websites, and I thought algorithmic bias was something important that they know about, and something important that their clients should know about, especially when clients were choosing whether to accept or not accept plea deals. I don’t know whether my writing made an impact on people’s lives. I only know my writing worked when it was fed into Google’s algorithm. It was good SEO. I hope it was also good for the world.
But that’s precisely the problem. What’s good for SEO isn’t always what’s good for the common good. What goes viral isn’t always moral nor is it always in the best interests of our culture. Anger and hate goes viral. Nuance, complexity, and poetry, not so much.
When it comes to racist algorithms vs. poetry, you already know which one is going to win. My question is, how do we change our engagement with the online world so that poetry and nuance win more? I think this is where critics come in. Which brings us back to gatekeepers, and our trust of gatekeepers.
But these days, the gatekeepers are more often more beholden to advertisers than the public’s best interest. And because of the free flow of information on the web, everyone can be a critic. Coded Bias touches the heart of a knowledge crisis in America and also the world. If algorithms are biased in favor of the wealthy and powerful, if people are biased because of these algorithmic feeds, and if capitalism demands that content sells, thus benefitting the capitalized and the powerful, who can we really trust. Who are our real critics? The next generation of critics might need to critique the web itself, particularly the more popular and influential spaces on the web. But who will trust them?
About the Writer
Janice Greenwood is a writer, surfer, and poet. She holds an M.F.A. in poetry and creative writing from Columbia University.