Recruiters and flexible-work platforms are adapting artificial intelligence to rank candidates and assign work. Unchecked, artificial intelligence will systematically discriminate, omit, and erase billions of people. To avoid it, companies and their technology providers should: 1) disclose when using AI, 2) make the factors transparent, 3) disclose rankings to candidates, 4) detect bias, and 5) reduce discrimination systematically.

Artificial intelligence is becoming a dominant force in recruiting. AI is not only comparing candidates and job opportunities but also matching candidates and teams. On flexible-work platforms, AI is already deciding who gets the job. Primarily due to sociocultural conditioning, humans are prone to either conscious or unconscious bias. Unconscious bias also affects the architects of AI. The individual contexts and upbringing of the architects shape their understanding of concepts like fairness and equality. In turn, this affects the algorithmic abstractions and models they use to represent the world. By far, most people who apply for a job don't get hired. What’s more, most candidates never find out why they weren't ’good enough.’ For them, the recruiting and decision-making process is an impenetrable black box. Unchecked, artificial intelligence adds further complexity to that impenetrability. Discrimination and inequality are becoming automated. Architects of algorithms are starting to hide behind closed doors. Unfairness in favor of a few will prevail to an even greater degree. It's a dangerous, slippery slope. Fortunately, we can also use technology to reduce bias. We can use behavioral design and transparency to promote equality (here is how the UK used these tools to increase the share of women in boardrooms from 12% to 25% in just four years). We can open the recruiting black box: teach candidates how decisions are made, detect discrimination and reduce it, provide automated and data-driven job advice on a massive scale, educate leaders about the dangers of runaway AI discrimination, and much more. Doing it won't be easy, but the first step is committing to it. Principles alone cannot guarantee ethical AI. Clearly defined, impactful pathways are needed. Similarly, AI ethics must be underwritten not only by individuals, but by businesses and organizations as well. At Torre, we support of the manifesto below. If you are a recruiter, company executive, a flexible work platform manager, a leader of an HR tech company or job board, or a policymaker, we invite you to support it, too.