AI revolution: the data scientist’s view
Our new AI revolution series focuses on how experts around Schroders are thinking about the fast-moving world of artificial intelligence. In this first Q&A, Parimal Patel, PhD, explains how his team of data scientists use data and AI to help inform investment decisions at Schroders.
Parimal, how does your team currently use AI?
We’re a team of data scientists working within investment at Schroders. What we don’t do is use AI to build models and algorithms to trade. Instead, we use the datasets we have and AI techniques to enhance our investors’ views so that they can make better investment decisions.
Rather than using AI to replace people, we use AI to provide an information edge in investment decisions.
What do you see as the limitations of AI?
This is still a relatively young technology and there will be leaps and bounds in progress in the years to come. However, as a group of AI practitioners, we would be quick to point out limitations and issues so as not to mislead anyone as to what can be done, and also what is advisable to do.
For example, the data required to go into a model needs to be complete and not full of gaps. Otherwise it will produce inaccurate or misleading outputs. Models themselves can throw up spurious connections. AI is not (yet!) magic.
What kind of AI techniques does your team use?
We typically use Natural Language Processing (NLP) and Machine Learning (ML).
NLP is used to save time – because reading through large amounts of text is time consuming – and to reveal insights that may otherwise have remained hidden.
ML encompasses a range of algorithms. These are again used to save time and gain insights by sorting through large datasets to produce predictions or recommendations based on historical data.
Do these AI techniques replace a fund manager?
Not for us. Our stance is that these approaches are best used to enhance a fund manager’s views: augmented intelligence rather than artificial intelligence.
Of course, there are funds in the industry that use algorithmic AI trading strategies but the world of investing isn’t bound by a clear set of rules and perfect data sets. The information and judgement needed is far beyond what currently exists in the AI world.
Given the interest in ChatGPT, what do you see as the future of AI in investment?
Large Language Models (LLMs) such as ChatGPT have tremendous potential to enhance productivity.
Another benefit, often unmentioned, is that the ease of using LLMs has sparked imaginations and reduced barriers to the adoption of AI techniques in general. Sceptics are being converted.
We see more acceptance of the use of sophisticated techniques to extract that information edge. For us, it will continue to be to support rather than replace the human decision.