In this pecha kucha presentation, Sean Gorman, CEO of Timbr.io, gave a high-level overview of how open algorithms and data ethics are connected.
Sean raised several important questions:
- As facial recognition software gets better, do we all just become barcodes?
- When an algorithm makes a bad decision, who is accountable: the developer, the data scientist, the company that ‘owns’ the algorithm, or someone/something else?
- What is classified as hate speech, and how can an algorithm identify hate speech?
Building on these questions, Sean posed this overarching question: Can open algorithms help us better understand data ethics?
In short, yes. As a first pass at answering this question, Sean explained that we must better understand how algorithms work. We know that every algorithm has bias. For example, algorithms that handle social media can lead to algorithmic racism, where an algorithm identified photos of people as animals, or algorithmic injury, where poor GPS data caused a four-car pile-up. If we know that algorithms have these types of bias, we can potentially address this bias through open algorithms in the following ways:
- We can learn from companies like Google, Microsoft, Facebook, which are experimenting with open algorithms
- “Real time notebooks” and dashboards can empower data scientists with better information
- We must better understand our own bias in order to better understand algorithms