Can you trust AI?


inCompliance magazine is the exclusive publication authored for the International Compliance Association. It goes out to their 13,000 members 6 times a year and the full magazine is exclusive to members of the ICA. Caspian were asked to contribute an article to the latest issue so Chris Brannigan and Joe Faith explored a challenging yet exciting topic for financial services – Can you trust AI?

AI is supporting high-risk, high-consequence human decisions across autonomous vehicles, healthcare and criminal justice whilst headlines estimate potential savings of $1trillion from AI by 2023. This opportunity also presents challenges in high risk environments such as financial crime and the potential of explainability sits at the core of the article.

You can read the whole article here with highlights including:

  • how machine learning is being applied to domain problems
  • how and why decisions are made
  • what happens when AI goes bad
  • explainable algorithms
  • the big role humans have to play

The potential for AI in high risk sectors is enormous and the progress made towards how machines are built, trained and assured is fascinating. Especially when you put humans in the loop.