The wrongful arrest and imprisonment of a Michigan man last year brought a kind of watershed moment to the tech industry. Facial recognition technology used to identify the man had misfired badly. The incident called into question not just the accuracy of facial recognition, but also its usefulness and purpose. Amazon, Microsoft and IBM quickly halted sales of the technology to law enforcement.
For MIT researcher, poet and computer scientist Joy Buolamwini, the story doesn’t end there. Her organization, the Algorithmic Justice League, takes aim at bias that exists across all technologies, not just the ones with obvious failure to accurately recognize people of color or women. But facial recognition technology is a particularly striking example, with an inaccuracy rate of up to 21% for gender and 19% for skin type, according to the Pilot Parliaments Benchmark study.
Systemic bias exists in the development of AI technologies, according to Buolamwini, who points to “largely homogeneous teams” that design the algorithms. AJL’s primary aim is to create “equitable and accountable AI,” done through outreach to corporations and the public, to raise awareness around the issues and provide information. The documentary film “Coded Bias” (see trailer below) follows Buolamwini’s ongoing effort to bring about legislation that governs algorithmic bias.
At Signal 2021, Buolamwini will share the extent of the issues in the technology, and how algorithmic bias can be overcome. Click here to reserve your spot for July 14-15, and watch the “Coded Bias” trailer below.
