In the glorious AI-assisted future, all decisions are objective and perfect, and there’s no such thing as biased decisions. That’s why we created AI and machine learning, right? Well, there’s some bad news: we humans are the ones building these tools, and as a result humanity’s biases and missteps can subtly work their way into the AI and models we make.

All hope isn’t lost, though! This talk will show you how to think through six categories of common problems, ranging from simply confusing causation and correlation to the pernicious effects of feedback in your algorithm, and gives you concrete steps you can take to avoid them, including some awesome tools you can use. The algorithms you write today could literally affect people’s lives, so you owe it to them and yourself to be able to deliver the best possible intelligence.