It’s commonly accepted that AI systems may have built-in biases that reflect those of their creators. How an AI decides whether or not to grant someone a loan, for example, may incorporate discriminatory prejudices. Too often, too many companies rush past measures that would have them carefully consider this problem. Fortunately, other organizations are stepping in.
One example is the AI Now Institute, which according to this recent article from MIT Technology review, will work toward identifying and highlighting instances of algorithmic bias. The institute is housed at New York University and was founded by Kate Crawford from Microsoft and Meredith Whittaker from Google. The team of researchers will partner with the American Civil Liberties Union on this initiative.
Another group cited in the article is O’Neil Online Risk Consulting & Algorithmic Auditing, which helps companies identify and fix biases in their algorithms. It’s led by Cathy O’Neil, a former Barnard professor and hedge fund quantitative analyst who authored Weapons of Math Destruction, a book which describes the danger of algorithmic bias in many use cases.
These organizations have a sizable task ahead of them, but admitting there’s a problem is always the first step.
Read the full story: Biased Algorithms Are Everywhere, and No One Seems to Care (MIT Technology Review)