It was a fairly typical summer evening in our neighborhood outside of Washington DC: I didn’t feel like cooking, so I took our 3 kids to a local pizza place for dinner.
As we squeezed into the booth, a friendly and talkative server took our drink order. As my 10 year- old daughter leaned over to grab some napkins, her long springy black curls bounced over her shoulders, like hundreds of pressurized mini mattress springs that had suddenly catapulted into freedom.
“Oh my gosh, her hair is so beautiful!” the server gushed as she ran her hands through my daughter’s hair. “Where is she from?”
“Where is she from?” might seem like a harmless question, and maybe it is when it’s directed towards a 10 year old. But is that same question harmless when it’s directed at an adult looking for a job? Is there a price to pay for acting, sounding or looking different enough that someone looks at you and wonders “Where are you from?”
This question has been studied a lot, and as it turns out, there is a price to pay for being different, and it isn’t cheap.
Vivienne Ming’s Work With AI and Bias In Hiring
Vivienne Ming, a theoretical neuroscientist and co-founder of Socos Labs, has a phrase for the price we pay when we have different hair, different names, different norms, and different backgrounds: it’s “a tax of being different”. The tax is largely implicit, Vivienne says, and people don’t need to be acting maliciously for the tax to be levied.
When Vivienne was Chief Scientist of HR Tech company Gild, she studied a data set of 122 million professional profiles, and found that one simple letter in a name could make a huge difference in a person’s job prospects: to be equally likely to get a promotion, someone named Jose would need a masters degree or higher, while someone named Joe would need no degree at all for that same promotion. Numerically, Jose would have to account for 6 additional years of schooling, along with missed opportunity costs. This tax would cost Jose $500,000- $1,000,000 over his lifetime.
That’s a pretty expensive tax for adding an “S” to the end of your name. And although Jose bears the direct burden of this tax, in many ways, we are all paying it. How? Decades of research by sociologists, psychologists and scientists shows that diverse groups (varying races, sexual orientations, genders and ages) are more creative, productive, and adapt at problem solving than homogenous groups. When we aren’t inclusive in hiring, we all lose.
Diversity pays, but the legal profession seems to keeps leaving money on the table. Why? Can we change our ways, and can artificial intelligence help?
Artificial Intelligence and its Impact on Diversity
The legal profession is known to be resistant to change, but it has slowly been adapting artificial intelligence in everything from automated document review to recruiting and hiring.
Diversity recruiting is one of the keys to sustainable success of a company, and artificial intelligence (AI) is indeed capable of recruiting, but what good are our AI systems if they are inheriting the biases of its programmers?
In Episode 27 of the Sweatours Legal Wellbeing Podcast, Vivienne states a fundamental fact about modern AI: if you do not know how to solve a problem, AI cannot solve the problem for you. In other words, if you hold an implicit bias in your hiring of women or minorities, AI won’t magically democratize the process.
There is hope though, because it is possible to train AI to ignore the things that don’t impact a lawyer’s job performance- things like sexual orientation, race, or gender.
Vivienne tells us that AI will reflect our own ethical choices right back at us, so arguably AI will make smarter decisions if we make smarter decisions first. By knowing something about the problem we are trying to solve, and by hiring human programmers who have engaged with real world problems, we can create, train, and refine artificial intelligence systems to be more diverse.
If we understand issues on a more human level and stop auto-prioritizing Joe over Jose, Aiden over Anjali, AI will follow suit. Subsequent inclusion of diverse employees, of course, will still be up to us.
AI is a phenomenally human technology, Vivienne states, if we choose to use it that way. In the coming years, the distinction between artificial intelligence and humans will continue to blur. As legal professionals, we need to ask ourselves: will we contribute to others’ sense of belonging , or will we destroy it?