Here's what I’m reading this week:
What Does AI Really Mean for the Financial Sector?: On Tuesday, White and Case published a detailed think piece on the implications of the expansion of artificial intelligence and machine learning within the financial sector. It’s a worthwhile read that addresses three key questions that will become more and more important as the uses of AI expand: (1) who owns the data that intelligent systems use to make their decisions? (2) how do we assign liability for machine decisions?, and (3) how can AI be used to help cybersecurity?
The most interesting section in the overview is the discussion of liability for machine-made decisions. The essential challenge is that as intelligent systems become more sophisticated, it is becoming harder and harder for mere mortals to understand why they make the decisions they do. If our goal is to build learning systems that solve problems we have trouble keeping up with, we’re succeeding – but it also means we’re out of the loop if something goes wrong.
This is a problem that we'll see grow larger across industries, but it will prove especially challenging in the financial sector for three reasons: regulation, complexity, and speed. First, the financial sector is heavily regulated, meaning that many decisions have some regulatory burden to demonstrate what was decided and why. Second, the inputs to financial decisions can be punishingly complex. This makes the environment fertile ground for intelligent systems, but also a hard space to keep up when the machines get moving. And third, trading systems move faster every day. There are already automated systems making trades in seconds, and keeping with that decision tree – or even unwinding a single decision when so many get made in moments – can prove incredibly difficult.
The writeup doesn’t offer solutions (and it would be surprising if it did, since most of what we have at this point are questions), but it points its focus at the right problems, and thats an important step.
I'm reading: "Three big questions about AI in financial services."
- Rob Knake and Richard Clarke share useful insights about what we need to do to prepare for the next time Russia looks to target our elections. The Politico piece offers sharp analysis, touching on everything from cybersecurity to influence operations.