EqualAI, a nonprofit organization and leading voice focused on reducing unconscious bias in the development and use of Artificial Intelligence (AI), today challenged corporations and business leaders around the globe to take The Pledge to Reduce Bias in AI.
Read More: Mi9 Retail Launches ThryveAI
There are immediate actions that organizations can take to identify and reduce bias within their existing systems. EqualAI has been seeking out ways that companies can go beyond mere statements in support of the Black Lives Matter (BLM) movement and commit to action. The Pledge identifies actions within a company’s control that they can commit to changing quickly to create a tangible impact in our society.
Current events make the stakes even higher:
- Terabytes of data are being collected around the world to track and evaluate the impacts of COVID-19. This data will be instrumental in creating a vaccine and implementing safety programs. But unconscious bias embedded in this data could further the disproportionate harms of COVID-19 in communities of color.
- The BLM movement is shining a spotlight on discriminatory practices, particularly within the criminal justice system. Bias in AI exacerbates discrimination and hides it behind the facade of a seemingly neutral scientific machine.
- As the global economy recovers, access to economic opportunity for underserved communities will be impacted by systems heavily reliant upon AI for data analysis and decision-making. We have already seen bias in AI impact hiring, lending, and other areas critical to a comprehensive recovery and cannot afford these harms to impede our recovery in the safest and most equitable manner.
Organizations taking The Pledge to Reduce Bias in AI are forward-thinking leaders. They recognize that addressing bias in the AI they use, create, or sell can protect and enhance outcomes for their customers, patients, employees, and society as a whole. Given the urgency, this requires bold commitments to change. The Pledge includes commitments to:
- Test and retest systems for unconscious bias in AI.
- Take appropriate, lawful action when harmful biases and patterns are identified.
- Ask vendors and partners to take The Pledge and commit to joining the effort to reduce harmful bias in AI.
“As we take this moment to assess our society’s treatment of those who are underrepresented and underserved — people of color, women, our elderly, people with special needs, and many others — we would be remiss if we failed to look at our technology and its role in sustaining disparities,” said Miriam Vogel, president and CEO of EqualAI. “We issued The Pledge to Reduce Bias in AI to provide companies with actionable steps to reinforce their public statements in support of Black Lives Matter and ending discrimination. While a pledge will not solve all of our challenges stemming from racism, discrimination, and bias, it is a positive and important step toward creating AI products that serve society and represent our values.”
Pymetrics, Humana, LivePerson, Postmates, and Cognizant are among the leading companies who are first to take The Pledge, which is supported by organizations committed to equality, including Catalyst and All Raise.
“Humana’s most important concern in using advanced analytics, including artificial intelligence, is to make health care experiences simpler for our members and their trusted care teams,” said Heather Cox, Humana’s chief of Digital Health & Analytics. “We hold ourselves to the strictest ethical standards and are proud to join our peers across industries to commit to responsible applications, fair and inclusive actions, trustworthy and transparent technology, and private and secure systems.”
“Bias in AI is insidious: it is hidden, buried in code, unintentionally added to products and services by people who don’t even know they’re doing it, and devastating in its impact,” said Robert LoCascio, founder and CEO of LivePerson, and co-founder of EqualAI. “AI harbors the unconscious biases that exist in our society, and unchecked, it can impact our opportunities — from our place and level of employment to our ability to obtain a loan — all without our even knowing it.”
Read More: How SMS Can Elevate Your Marketing Programs
“pymetrics is proud to take the EqualAI Pledge, as its principles are core to our mission of realizing everyone’s potential by matching each person to their best-fit job in a fair and ethical manner,” said Frida Polli, CEO and co-founder, pymetrics. “As more and more companies commit to and live by this approach, we believe we can reset the expectation that AI can and should be used for the good of individuals, businesses, and our communities.”
“From matching algorithms that support the growth of mom and pop bakeries, to apps that empower workers to earn income, and to automated delivery devices like SERVE — Postmates is committed to building equity in our systems from the ground up,” said Vikrum Aiyer, Vice President of Public Policy and Strategic Communications for Postmates. “Our pledge to the principles of EqualAI is born out of our mission to recognize that our code, our products, and our vision must continue to be fine tuned to address and diminish structural biases present in technology, communities and in any synergies in between.”
The Pledge to Reduce Bias in AI can be found on EqualAI’s website as well as below. In addition to taking The Pledge, organizations and leaders can:
- Take action: See the EqualAI Framework and EqualAI Checklist to Identify Bias in AI for guidance on best practices and how to start testing your AI.
- Partner with us: Support EqualAI and other organizations trying to end discrimination.
- Contact us: Share feedback on ways we can collaborate to reduce bias in AI.
EqualAI’s Pledge to Reduce Bias in AI
AI often acts as a mirror, reflecting and magnifying the values in our society — including our biases. Bias can enter AI through the human touchpoints throughout a product’s lifecycle — from a lack of diverse perspectives in the system’s design and development to the bias baked into the data from which it learns and iterates. And intentional or not, without our intervention, decades of fighting for equality can be unwritten in a few lines of code.
Through this Pledge, we commit to creating a better, more just reflection. AI may not be the sole source of our challenges with discrimination, but we will strive to use AI as a tool to reduce harmful bias, not replicate and spread it. We are committed to creating and supporting AI that is designed to serve and protect a broader cross-section of our population — and we will ask our partners and vendors to do the same.
Together, we can reduce the impact of harmful bias in AI. We hereby commit to fostering more equitable and inclusive AI by adhering to best practices, such as the EqualAI Framework and specifically:
- I pledge to test (and regularly retest) for unconscious bias in the AI that my company and/or I use or create, particularly in critical areas where we do not want to replicate and scale historical biases against protected classes, such as in hiring, evaluations, promotions, product testing, and awarding benefits (e.g., financial and health care services).
- I pledge that I will take appropriate, lawful action within my control when I identify harmful biases and patterns in AI that create inequality.
- I pledge to ask our vendors and partners to take the Pledge and commit to joining our effort to reduce harmful bias in AI.
Read More: Baidu Tops China’s AI Public Cloud Services Market, According To IDC Report