Technology is only neutral if built to be neutral. People build non-neutral technology all the time, either intentionally or unintentionally. If you take a hug

Technology is only neutral if built to be neutral. People build non-neutral technology all the time, either intentionally or unintentionally. If you take a huge random data set and train an algorithm off it, the algorithm is going to reflect the biases of the data, even though the algorithm - and the technology used to run the algorithm - is completely neutral in theory. If you accept this as a problem and try to solve it, you're taking an explicit stance of non-neutrality - something is wrong with the output of this algorithm based on my criteria, regardless of whether it is implemented correctly. For many real-world situations, neutral technology is the opposite of what is required to solve problems. Which is a real bummer, since it means that you have to make difficult decisions in addition to building a good piece of technology. Governments are already using "neutral" technology to make important decisions about people's lives [1]. But how neutral is it, really, if the data we feed into it for training is biased or incorrect? [1] https://www.nytimes.com/2016/06/23/us/backlash-in-wisconsin-against-using-data-to-foretell-defendants-futures.html https://www.nytimes.com/2016/06/23/us/backlash-in-wisconsin-...