A biased industry, unchecked
There is not much oversight (actually no federal oversight at all) right now, even though groups are fighting to change this. Artificial intelligence (AI) and machine learning are getting more powerful, to the point where the programmers who created AI do not know what is going on inside—a black box, if you will.
Still, those who control the lines of code have immense power, much of it currently unseen by the very people it negatively affects. Coded Bias describes how companies “test” a new code or process on communities of color before releasing it to the larger population.
History is messy but the huge amounts of historical data fed into the AI systems create a certain bias toward white people and against people of color. Dr. Buolamwini presents specific examples, including how Amazon’s AI hiring system is biased against women, and how people of color are targeted by police…for being non-white.
Bias in facial recognition technology
Our own faces may be the final frontier in tech, and companies are racing to conquer that frontier. Facial recognition infringes on privacy and individuality. The data and AI create a way for companies to pinpoint, with scary accuracy, how consumers (you and I) will react to certain ads that we see in our Facebook (or other social media) feeds.
AI is not always correct, and the movie shows how computer efficiency leaves no room for compassion and humanity. Don’t think that big business isn’t chomping at the bit to determine how you react to the fabulous products and services they offer. It’s still about control. But “there are forces fighting through ‘the ballot’ (not the bullet, yet) to reverse this unsettling trend with technology, privacy, and equity.“
Dr. Buolamwini herself created the Algorithmic Justice League to combat the efforts of biased data and AI systems. Along with watchdog groups all over the world she will be following developments of artificial intelligence.
Matt Crichton is a frequent contributor to WIP.
More information: Coded Bias, Algorithmic Justice Leauge, Gender Shades