Artificial Intelligence and investing
Will 7IM become 7 'AI' M?
In November 2022, the world was introduced to ChatGPT.1 A few months later, 100 million people are using the AI chatbot to answer questions, solve problems and, for some reason, write vast amounts of poetry involving everyday items.
Another thing people have taken to doing is ask Chat-GPT to build investment portfolios. As with lots of the service industries, it’s a natural response. After all, the main input into investing (and law, accountancy, auditing, consulting, etc.) is the human brain – so seeing if the artificial brain can have a go makes sense.
And the results are… okay.
The various portfolios ChatGPT spits out are reasonable approaches to long-term asset allocation. Broadly diversified, plenty of growth assets, a few defensive ones, maybe the odd sprinkling of gold or property.
Not a bad start – at the same time though, not game-changing. Of course, it’s early days. Bloomberg have just announced their GPT version, trained on billions of market data points, and aimed specifically at finance.
New technological development is exciting. It got us thinking about how using AI could improve our investment process. We’ve always believed that successful long-term investing needs a blend of art and science, with a healthy dose of psychology layered on top. There’s potential in all three areas.
Our fundamental investment framework is based on statistics. We spend a lot of time reading the latest methodological research, looking for longer-term datasets, and thinking about procedural improvements, to identify the right long-term blend of assets for any given risk profile.
At all stages of that process, AI can help us get better. If you’ve ever read an academic paper written by mathematicians, you’ll understand why asking a nonhuman to read and summarise it might be better for everyone!2
And when looking at new datasets, or applying new methods, AI can be extremely helpful in having a first go at analysis. It can write code, apply it, and then display the results – which allows us to think about the answers, and then ask different or more detailed questions.
Being able to repeat and enhance our scientific methods quickly (and without the very human frustration that comes with doing repetitive tasks) should allow us to take deep research insights and apply them far more efficiently.
We’ve always believed that investing requires art as well as science. The past can’t tell you everything – the future is created rather than predicted, and things can change in unexpected ways at unanticipated speeds.
But all creative processes need a starting point. Michelangelo didn’t just pick up a chisel and head for the nearest block of marble. Beethoven needed piano lessons, and Shakespeare took his inspiration from history. Art requires just as much research and planning as financial modelling – so there’s certainly room for assistance.
When thinking about a new potential investment opportunity, why not ask an AI assistant to provide a general summary of the current state of the world, or collate predictions, or form likely scenarios? The act of gathering research isn’t the value that an artist/investor can add – it’s what they do with that research.
Understanding psychology is crucial for successful investing. Without humanity, there are no markets. So whether it’s understanding limits of our own rationality when it comes to decision-making, or recognising the emotional responses of other investors, the human element is unavoidable.
This is the most exciting aspect of working with AI. In most scientific analysis, research takes place from an external perspective. We study the stars, or atomic reactions, or ecosystems from the outside. And every time we find a way to get a new perspective, we learn. A better telescope, or hadron collider, or undiscovered species always tells us something!
But with psychology, we are the subjects and the researchers! It’s unavoidable. The brain is trying to understand and analyse itself. There’s a distinct possibility that a different cognitive approach might tell us something about ourselves that we’ve never considered.
How that interacts with investment is too soon to tell – after all, behavioural economics (which incorporates psychology) has only been on the scene since the 1970s. There’s a lot of scope for discovery!
More than just a tool
We need to change the way we think about AI if we want to interact properly and get as much as possible from it. Using AI isn’t like having a software update on your PC, or a more powerful drill, or a more efficient battery.
Interactions with AI are going to be more like having a new colleague; someone who has a completely different skillset and different method of thinking to you. This could happen sooner than most people expect.
At 7IM, we have an investment environment where we try as hard as possible to encourage diversity of thought. We want to bring as many different, well-informed perspectives as possible to everything we do. Having a non-human perspective as one of those might just deliver something unique…
1 GPT stands for “Generative Pre-trained Transformer”
2 I did study maths, so I’m not just taking pot-shots!