Fooled by Randomness had a great impact in the way I think about all models and statistics. It has some really strong examples and cases that stuck with me. It's one of the more general.

If you're currently building models that work with personal and/or demographic data - things that can affect people - have a look at your team and think about the biases. If your team is dominated by a certain point of view (all men, all from one country, etc.) then maybe look at Weapons of Math Destruction or Invisible Women.

I hope that helps.

Dr Adam Sroka, Head of Machine Learning Engineering at Origami Energy, is an experienced data and AI leader helping organisations unlock value from data.