The topic of artificial intelligence (AI) has been around for some time now. We’re beginning to hear a lot more about it as many industries have begun to spend resources on it in a serious way. Looking back the past few years, two experiments in particular stand out as being the catalysts for the renewed interest in AI or smart machines. First, IBM’s Watson supercomputer won against two of Jeopardy’s greatest champions. More recently, Google’s AlphaGo program handily beat the world Go (a complicated board game originated in China) champion. Of course, many big firms are jumping into developing self-driving cars (where AI is a must) – Tesla, Uber, Google, and even China’s Baidu among them.
What Watson and AlphaGo were able to achieve is nothing short of amazing. Without the machine’s ability to learn, what they did would have been impossible. The general consensus view has been that technology in the next-so-many years is going to automate a lot of mundane tasks that are currently being done by humans, even in the areas of medicine, law, and investment management.
Many professional knowledge workers aren’t too worried that machines will be replacing them anytime soon as many believe they will still be needed to do the complex thinking. Unfortunately, a number of recent studies have shown that most human beings tend to perform sub optimally even in those complex-thinking areas. According to Professor Ed Hess of the University of Virginia, to compete and add value in the age of smart machines, most of us will need to take our cognitive skills to a much higher level.
According to an Oxford University poll of AI experts, full AI (machines reaching human level intelligence) has a 50% probability of being reached by 2050, escalating to 90% probability by 2075.
Let’s examine the potential impact of AI on the investment management industry, in particular. Technology (algorithms, more specifically) is nothing new to the investment industry; most hedge fund shops and trading systems have benefited from it (speed and accuracy have been the key). AI is the next natural phase, using machines with the ability to learn and adapt.
First, let us examine the pros and cons of AI.
- With AI, the chances of errors are almost zero on routine, rules-based tasks
- Emotions that often hinder rational thinking for a human being are not a hindrance for A
- Smart machines can crunch vast amounts of data very efficiently and continuously
- The cost to program and maintain AI is very high, at least at the present time
- Smart machines lack the so-called emotional intelligence and intuition of humans – these human attributes have been shown to play an important role in certain situations
- With the heavy application of AI, humans may become overly dependent on machines, diminishing mental capacities
Already, some very large AI systems are deployed by well-known hedge funds to capitalize on the latest advances in deep learning. Renaissance Technologies is a prime example. One of its hedge funds is being co-managed by Peter Brown and Bob Mercer, who developed natural language recognition programs at IBM. There are others, including Bridgewater Associates, Two Sigma, BlackRock, and Goldman Sachs, that have begun to deploy a lot of resources revamping their AI-driven systems.1
With advances in technology, certain investment-related jobs (that were once handled by humans) are beginning to disappear. For example, traditional junior analyst tasks, such as financial statement analysis or industry trend analysis, can now be done by AI in real time, at a fraction of the cost. According to David Kedmey, CEO of EidoSearch, the legacy financial analyst position, which sifts through SEC filings and press releases, is already disappearing.
It’s not all bad news. In this environment of increasing automation, people with certain types of skills and knowledge are likely to become more valuable. For example, data scientists will be in high demand to devise the algorithms that will fuel future AIs. In addition, according to John Rubino, a financial writer, we’ll be needing more “hand holders.” For a vast majority of institutional investors, investment performance isn’t as important as being able to easily understand the underlying investment strategy. Hence, investment management firms will likely need more articulate and empathetic people (hand holders) to work with investors, which probably can’t be automated.
In many ways, AI still has a long way to go in terms of matching human intelligence. According to an Oxford University poll of AI experts, full AI (machines reaching human level intelligence) has a 50% probability of being reached by 2050, escalating to 90% probability by 2075.2
I don’t believe we need to worry about smart machines replacing every one of us, whether it is in the investment management industry or elsewhere. The most probable scenario is it will be more of a collaborative environment between machines and humans. Humans are simply not as good as machines when it comes to certain tasks – crunching/analyzing vast sums of data with accuracy and speed, for example. Furthermore, most human beings have what psychologists call cognitive biases – a systematic error in thinking that affects our judgment or decision-making skills. Smart machines shouldn’t have these cognitive biases, at least in theory.
As we look ahead, those investment management firms that are best able to incorporate many key aspects of AI could dominate the industry. It’s not an accident that firms such as BlackRock, Goldman Sachs, and JPMorgan are beginning to put serious resources into revamping their AI-based systems.
1 Integrity Research Associates
2 Future Progress in Artificial Intelligence: A Survey of Expert Opinion