The report describes how Algorithms and Artificial Intelligence bring the potential for widespread discrimination
As algorithms and AI become ubiquitous in our modern world, their capacity to make life better becomes clearer - as do many critical risks. Open MIC's new report, “Formulas for Trouble: Why Smart Companies Must Tread Carefully With Algorithms,” highlights the need for businesses to beware of baked-in bias.
AI is seen by many as the most exciting and fastest growing technology of the 21st century. Experts predict that it will help cure diseases and address global challenges such as climate change and world hunger. At the same time, alarms are sounding about possible negative societal impacts of these data-driven tools.
The report, released today by Open MIC, a non-profit that works to foster greater corporate accountability in the media and tech sectors, highlights the potential for such abuses – from Facebook’s algorithms allowing companies to exclude older workers from job ads, predictive policing technology leading to further surveillance of already over-policed communities, or Apple’s customers discovering that the iPhone X is programmed to recognize white faces, but has repeatedly failed to recognize the faces of people of color. Even when algorithms are intended to counteract discrimination, bias can still occur.
“Companies that use these technologies must have transparent oversight, and they must take into account the impact that algorithms can have on people’s lives,” said Michael Connor, executive director of Open MIC, which works to foster shareholder engagement with media and tech companies. “Investors can and should play a key role in encouraging companies to identify and address the risks of these technologies.”
“Life and business in the algorithmic 21st century must be developed around the core principles of fairness, accountability and transparency,” says the Open MIC report. “Companies in every industry sector — especially their senior executives and boards of directors — will need singular focus and commitment to those principles if the enormous promise of artificial intelligence is to be realized.”
Technology experts, corporate leaders and other specialists offer mixed responses on whether algorithms and AI will ultimately help more than they hurt. Risk assessment tools are a critical first step, but all agree that significant investment is needed to identify better policies and practices to monitor the potential for bias.
The importance of data-driven AI tools was highlighted this week in hearings on Capitol Hill, where Facebook CEO Mark Zuckerberg repeatedly told members of Congress that he viewed AI as a solution to dealing with many of the company’s content problems. But when pushed on the potential pitfalls of such technology, Zuckerberg admitted, “Right now, a lot of our systems make decisions in ways that people don't really understand.” He said it was “extremely important” that AI tools be developed with transparency in mind.
Michael Connor called for Zuckerberg to honor that promise, and for other CEOs to step up, saying: “Responsible businesses that are open to promoting long-term, sustainable and equitable economic solutions have the opportunity to be leaders in their industries. It’s time to for companies - and their senior management and boards - to get serious about addressing potential discrimination, or risk a new standardization of bias that is more serious than ever before.”
Read the full report “Formulas for Trouble: Why Smart Companies Must Tread Carefully With Algorithms,” here: http://formulasfortrouble.openmic.org/.