As with many investor conversations these days, this one quickly turned to AI. A fund manager I spoke with recently left to put it well: “A car doesn’t drive itself; you need to go or at least tell where to go. But if you want to get to Manchester, it’s a lot easier by car than by foot.’
And this is the heart of the question of artificial intelligence for high-value-added intellectual activities such as investment. It’s not really about whether AI can do it better than a human, but how AI can help a human do it faster, more accurately, or at a lower cost.
The lawyer added at the same lunch that when he started out, a key part of an intern’s job was “discovery” — sifting through reams of documents to find information their boss could interpret.
The same in investments, where, strangely it seems now, the company I used to work for had the watchword “information advantage”. This meant a large team of numerous but relatively unskilled young people sifting through annual reports to populate a database with financial ratios, which in turn allowed investors to do the real work of picking stocks.
One of the most interesting potential uses of artificial intelligence is in expensive and resource-intensive investment analysis. Can machines really do it better? The answer is yes and no.
The reality of this “information advantage” was that it was the difference between some pretty basic knowledge and none at all. The effort that went into creating this fringe benefit was substantial, expensive, and time-consuming. It’s the work that AI promises to do in a fraction of the time at a fraction of the cost. Like the thousands of people who worked with horses in 19th-century London or New York, these datagrunts will have to find a new way to make a living.
So AI is potentially a labor-saving device – a multi-purpose vacuum cleaner or dishwasher. Another way it could improve our lives as investors is by removing many of the unhelpful behavioral biases that cloud our ability to make good decisions. People are generally more interested in stories than data. AI doesn’t care.
A recent study of artificial intelligence and stock selection from the University of Chicago Business School tested whether the revolution could move further up the food chain than the data collection that gave my former employer an informational advantage. They sought to see if artificial intelligence could take over from analysts who want to turn basic financial data into predictions about the future direction of corporate profits.
This is a key part of the investment process for active fund managers, providing the raw material for stock selection. The study found that machines did a little better.
More interestingly, he found that portfolios constructed using these earnings forecasts outperformed in a statistically significant manner. It’s still early days, but jobs at risk across a range of professional environments are getting older – and alarmingly fast.
One danger of artificial intelligence is that it makes the complex business of managing our money seem easier than it probably is. Ask an AI tool a simple question and it will magically come back with an answer that looks like your silver bullet. Want 10 income stocks with a dividend yield higher than a cash fund? Here you go. A portfolio of growth stocks trading at less than 10 times expected earnings? Try these.
Actually, this is nothing new. Thirty years ago, the first screening software appeared that allowed investors to plug in a set of criteria and spit out a shortlist of potential investments. The problem for most of us was the price of these services. AI democratizes the process. But it also turns it into a black box.
I trusted the numbers my Online REFS inventory screening software relied on because I saw my young colleagues inputting them into the system. When I ask ChatGPT a question, I have absolutely no idea where to go to find the answer.
AI plays on our penchant for shortcuts. We want this list of 10 stocks to be the answer, so it’s tempting not to ask too many questions. But as they said in the early days of computing: garbage in, garbage out.
I’m excited about what AI promises investors, but I’m also relieved that it came at the end of my career and not the beginning. Many of the things I’ve been paid for over the past 35 years or so have little value today. To continue the horse analogy, I adjusted a lot of bridles and cleaned out a lot of stables. Now I have to understand internal combustion or step aside.
However, I’m sure there will always be things people do better. We are social animals. I want the investors managing my money to be AI but human. I want my financial advisor to ask about my children. I don’t just want the car, I want someone to show me how to drive it.
The AI ​​revolution is at least as disruptive as the internet boom that preceded it. It will be divisive. This is great news for the small number of people who are already highly skilled and highly paid. It will be painful for a much larger number whose contribution is no longer so highly valued. Winner takes all is not a recipe for social cohesion. But it doesn’t go away.
Tom Stevenson is Chief Investment Officer at Fidelity International. Opinions are his own