I set up this blog back when I was hedge fund portfolio manager using machine learning algos to trade credit. These days I advise (mainly banks) on using AI & machine learning tech, as well as architecting and developing AI applications for them.
So I’m out in the field working with this stuff every day, trying to figure out what to do with it so that my clients make more money.
In my previous life as a trader, I encountered many non-trivial issues relating to information theory and the limitations of inference based on historical data in a strategic game.
The game I focussed on was financial markets, but a lot of human behaviour, be it in commerce, global politics or even competition for sexual partners can all be considered a strategic game. At some point ML/AI will encroach a lot further into these areas.
Tackling these problems forced me to think more deeply about the limits of machine learning, how it it differs from human intuition and strategic thinking, and where development in areas such a deep learning may lead. I’m surprised by the lack of sensible public debate about such an important technology, so thought I’d thought I make a space that would force me to order my own thoughts and perhaps spark some discussion.
P.S. I realise there are many smart people writing about these issues form a technical computer science and software engineering viewpoint. I want to take a more general behavioural/psychological/game theoretic approach because this is something far greater than an engineering problem.
Disclaimer: I know this is a cliche but I had probably better say it given the number of financial market related posts: nothing on this blog constitutes investment advice.