Ask a person on the street, and chances are they’ll tell you they are both optimistic and anxious about AI. The conflicted perspective makes sense—AI is already appearing in ways that have the potential to both scare and inspire us. The 2018 Fjord trends for business, technology, and design suggest a potential path to alleviate those fears: Adopt a values-sensitive framework for Responsible AI.
The good, the bad, and AI
Thus far, AI optimists have had much to celebrate. Maybe they have it easy: Intuitively, we associate technology with progress—we generally believe that “better, faster, smarter” leads to improvement, efficiency, and enrichment. In many ways, AI exemplifies those qualities. And as intelligent machines learn how best to pair with human savvy, it’s easy to picture new, improved ways of working. From the creation of jobs to more engaging work, we’re just beginning to see the fruits of those efforts, and the results are exciting.
But technology is not a neutral force. As AI plays a bigger role in systems that affect social outcomes—like criminal justice, education, hiring, or health care—it’s clear that the creation and shape of AI decision-making needs to be taken seriously. What happens when algorithms decide whether or not you get a job, home, or loan?
An opaque “black box” that decides who gets what and when is troublesome. With this new algorithmic power should come great responsibility. At a minimum it is critical that intelligent machines be able to explain their decisions as well as offer a sense of how they will make them going forward.
How can we ensure that AI in everything from lending to household goods is designed to be fair, transparent, and human centric? The Responsible AI movement is tackling those questions by proposing a framework that can help organizations foster AI with principles. Today, three tech questions make it clear that the movement’s work is addressing issues critical to our lives now:
1) Why do I feel like I’m being watched?
When you add the proliferation of camera-ready devices and exponential progress in AI, you get computers that are visual learners. From facial recognition payments to tracking babies as they sleep, the possible applications of machine learning are widespread. But while troves of visual data are being turned into compelling insights, there remains a challenge of how AI can collect visual data responsibly.
Computers still need to get better at reading human emotions and behaviors. But organizations will need to decide sooner when a visual input warrants a computer response; it’s important that machines learn to respond in a way that puts people at ease. What’s more, firms will need to decide what kind of data should be extracted: when gathering data look for visual information that’s useful and whose collection doesn’t ruffle feathers, or laws.
2) AI knows you better than you do
AI may soon know our preferences better than ourselves. Smarter algorithms are propelling chatbots, voice, and messaging platforms to drive today’s retail discoveries. But guess what’s left out? Branding. Audience-targeting algorithms lessen the impact branding efforts, endorsements or PR campaigns. And because brick-and-mortar stores aren’t as ubiquitous as they have been in the past, AI recommendations sit between logistics operations and customers.
Brands will need to tread past these gatekeepers carefully—crossing the moat of e-commerce platforms without losing customer trust is key. It’s easy to worry that your intelligent assistant has too much power—including the power to manipulate and embarrass. And with AI looking to connect at home, in the car, or anywhere you take your location-tracking phone, the possibility for fatigue is real. Further, when AI can be gamed or hacked to limit choice, a shared set of best practices that addresses authenticity and manipulation is critical.
3) Your new algorithmic coworker
Worries about the rise of the robots don’t totally capture AI’s role in the future of the workplace. Yes, jobs will be lost, but new jobs will take their place, and better ways of working will emerge when AI is paired with humans—not competing against them.
Responsible AI principles will be crucial for machine and human coexistence at the office, however. We’ll need to consider the specific types of job training and counseling programs that help people adapt. What’s more, guiding AI toward collaborative projects—all the while considering its needs as “another type of user”—will require designing machines to ask the right questions and reduce ambiguity. Algorithmic transparency is particularly important to this endeavor. AI bias reflects the same biases that humans have—so it’s critical that a diverse array of people, data and machines work to keep biases in check.
As these trends show, it’s essential we make sure AI is socially accountable. And with AI’s rapid incorporation in so many aspects of our lives, now is the time to be addressing how it is working and how it isn’t.
Not tomorrow. Today.
This article was produced on behalf of Accenture by Quartz Creative and not by the Quartz editorial staff.
