In the run up to the second Nextgen Alpha Artificial Intelligence Investor Conference in Frankfurt on 26th October, now seems a good time to thank all those who responded to the first AI Diary that I sent out in early August, and to summarise where we are prior to the live discussion. It’s also an opportunity to slip in some more ideas gleaned from the changing world around us.
The first AI Diary was a series of questions. Summarising the answers received:
• Does the fund management industry need AI?
The answer to this is yes. Every other sector of the economy is embracing AI, so why should fund management be any different? From healthcare to agriculture, human experience is being augmented by the equivalent of millions of years of collective experience courtesy of machine learning. By the way, robo-advisers were dismissed as doing no more than automating the past and nothing to do with AI or the future.
• Are machines alone good enough to win without human intervention?
Answers to this question were less clear cut, suggesting that a hybrid mixture of artificial intelligence and human intelligence (HI) is likely to remain the dominant operating model for some time to come. Automating as many processes in the value chain as possible seems sensible with research focusing on how to take the next step and then the one after.
• What happens if inflation makes a comeback widening the gap between real and nominal returns?
The consensus that emerged is that machine learning can cope with this scenario along with others that require non-linear multi-dimensional models.
• What happens if listed markets fail to capture growth opportunities because winning companies stay private?
There was no answer to this one. It’s a problem for conventional fund management, whether based on HI, AI or a combination. An FT editorial linked the move away from public markets as widening inequality and weakening democracy. Strong stuff.
• What happens when governments change the rules for non-financial reasons?
Once again no real answers, but as above, this is a problem for all investors. Whether HI will be quicker to spot increasing political risk than a machine is yet to be tested.
• Will AI change the cost of capital and have an impact on volatility?
Undoubtedly, just as the growing importance of ETFs is doing at the moment. Size matters in financial markets and the dominant players make the rules.
Unasked in the first note, but raised by several of those who responded is that AI will improve the decision-making process by removing emotion. Here, I take issue with the idea that emotion is a bad thing. In the world of sport, champions not only prepare meticulously, but also manage to channel all their emotions, plus an overdose of adrenalin on the right day, and at exactly the right time. Those with a tendency to panic lose. Why should the investment world be any different? Of course, practitioners who never put themselves on the line are likely to get it wrong at times of stress, but some feed off uncertainty and make exceptional decisions. I can see how you can model average behaviour and do better, but modelling the best seems less straightforward.
If AI starts to dominate fund management because of good results, people will quite reasonably surrender control to machines. The problem with this is that people will stop learning how to make decisions. Fund managers make decisions every day, some good, some bad and the good ones learn from their mistakes. The US military used to send senior officers to Wall Street to play investment games in order to improve their decision-making capabilities. Their day job demands that they make just a few decisions, usually at times of extreme stress, and it is better that they have practised away from the sound of gunfire.
Talking to a CTA manager a few years after the credit crunch, what struck me was the sense of helplessness when markets exposed flaws in the system. In answer to the question, what did you do when the fund went down 10% in a month? The answer was, “we reviewed all of the decisions made, we tested out models and…… hoped for the best”. Avoiding catastrophic loss may require human intervention. Equally, it could be right to leave the machine to sort out the problem, find the bottom of the valley and come up the other side. Keynes’ observation from a different era resonates when he said that, “markets can remain irrational longer than I can remain solvent”.
The great advantage of developing AI investment management strategies is that they allow us to be our very best 24 hours a day, 365 days a year in every market. The trouble is that systematic strategies are based on judgements made by fallible humans and so will eventually fail. If AI can overcome human failings, then its game on. That is until a few winners come to dominate the market making the rest of us redundant, retired or redeployed. Then perhaps this is no more than an emotional reaction caused by reading The Fear Index by Robert Harris this summer.
Investors should remember that the value of investments, and the income from them, can go down as well as up. Investors may not recover what they invest. Past performance is no guarantee of future results.
Any mention of a specific security should not be interpreted as a solicitation to buy or sell a specific security.