媒体

Third Bridge Views: Busting the Bot Myth: Why the Investing World Still Needs Humans

M

In this edition of “Views from Our Executive Team,” Third Bridge co-founder and CEO Emmanuel Tahar scrutinises the myth that Artificial Intelligence can fully replace human insight, particularly in the investing world

The so-called “March of the Robots” has totally captivated the public’s mind, especially over the last year. Terms and technologies like Big Data and Artificial Intelligence (AI) have taken the worlds of industry and finance by storm: their potential is no longer just the talk of tech-savvy experts, but of global leaders and business moguls, too.

There’s something about this year that seems to be making people lose their minds over these disrupting technologies. There is a well-established cycle that many of these mini-revolutions go through: a concept gets dreamt up and developed, then it soars into the limelight as ‘the next big thing’ before being deflated as reality fails to live up to expectations. Gartner’s Hype Cycle for Emerging Technologies perfectly captures this trend in the business world:

Image result for gartner's hype cycle 2016

It wouldn’t be fair to call Big Data or AI ‘buzzwords’ anymore – but could the finance industry be in the midst of a hype cycle when it comes to their potential to revolutionise the world of investment and research? It seems so, and this is especially true in looking at how automation is breaking out of the factory and into the office, whipping up traditionally high-skilled sectors into a frenzy as algorithms seem to be outpacing and undercutting the human mind.

Getting Carried Away

The World Economic Forum (WEF) in Davos has hosted hours of discussions on how to handle unprecedented rates of automation. Kai-fu Lee, the influential tech venture capitalist with 50 million social media followers in China, is confident that “pretty much anything that requires 10 seconds of thinking or less can soon be done by AI or other algorithms.” Elsewhere, WEF founder Klaus Schwab told Sergei Brin how a prime minister of “quite an important country” told him “there are three powers left in the world – one is the US, one is China and one is Alphabet.”

These world leaders aren’t getting excited because they’re tech geeks, though. Anyone with a passing interest knows that Big Data’s potential has been well-appreciated since the Noughties, and automation has been replacing people for decades. The two fields were always on a collision course, but it’s only now that the sparks are flying.

Big Data used to be seen as a headache by both the science and business communities. The only question anyone had time to ask was “how do we handle it,” as opposed to “how do we make the best use of it?” 

AI was, for a long time, confined to replicating processes that humans already fully grasped, with machine learning too raw to start taking advantage of information that we don’t already know how to handle.

Collision Course

In fairness, we still haven’t got very far. The difference now is that, in many industries, information technology has tipped the balance between opportunity and cost enough to spur large-scale changes in strategy. No sector offers a better example of this than banking and finance.

Improvements in data processing and trading algorithms seem to be turning the industry on its head, making quantitative trading, investment and research strategies look more worthwhile. Investors have started thinking more critically about the fees they pay to fund managers.

This has plunged companies with human-intensive approaches into difficulties. Paul Tudor Jones’ hedge fund, for example, had to lay off about 15% of its workforce last year amidst $2bn of investor withdrawals. Firms that can afford it are shifting from active to passive management, as exemplified by BlackRock’s massive overhaul, announced in March, which resulted in the addition of nine quantitative-strategy (quant) funds and the phasing out of some traditional stock-pickers in a move affecting some $30bn of assets.

Man vs Machine

It’s too simplistic to say that the quantitative is winning over the qualitative. BlackRock’s quant funds have, arguably, yet to prove themselves. Almost two-thirds of their quant lineup underperformed last year, with four-of-their-five main quant hedge fund strategies seeing losses. The firm was obviously confident enough to start their robo-revamp last month, but with an emphasis on the value proposition of active managers – whether their high fees were worth it.

Hedge funds, while still getting returns and posting a collective 2.63% in the first quarter of this year, are just not doing as well as the indices, which passive funds track. The S&P 500, for example, saw a 6.07% return in the same period.

It’s simply too early to call a winner in the fight between man and machine. Indeed, in many areas of finance, there probably won’t ever be a clear victor. Data is powerful, but it also has limitations. Humans have their limitations too, but have power in other areas beyond the biggest and most sophisticated mainframes.

The Limits of Data

Of all the information that’s indexed, we only know how to process some of it. How that information is processed and what gets done with it will always come down to human decision-making.

This is why even the most ‘robo’ of robo-advisers still have finance experts leading their algorithms’ developers. The same goes for other sectors: JPMorgan’s famous COIN machine, for example, still requires the oversight of lawyers and loan experts.

Roles like these are valuable, but the debate over AI and Big Data often gets carried away with them as well. World leaders are worried about returns from technological advances going to those who own and run these machines – but this ignores all the information that can’t be indexed by computers or of which only people can make sense.

After all, of all the information out there, only some of it is indexable. Out of what is indexable, only some of it is indexed. This limitation is exactly why clients have come to Third Bridge looking for answers as they consider investment opportunities ranging from a Kenyan oil field left abandoned since 1974 to the publishing industry in Kazakhstan to the market for instant noodles in Nigeria.

It’s within this information space where a role almost entirely for human-driven approaches exists. Again, the AI debate gets carried away with this: the typical argument goes that only emotion-based roles are truly safe from automation in the long run, and that doesn’t sound very relevant to the world of investment. Interpersonal relationships and human experiences cover far more than that.

People Power

Human expertise allows us to make links that machines cannot. Face-to-face interactions can get us to the source quickly and efficiently. This covers far more than obscure market research, too. We can take decisive action when shocks happen in well-researched industries and make sense of them while financial algorithms freak out. While stock-picking algorithms are good at handling day-to-day financial developments like earnings reports and commercial orders, they’re often useless when the stakes are high and the unexpected happens.

Take the Wells Fargo scandal, for example. Following an investigation last September, the bank was found to have been fraudulently opening fake accounts for customers, resulting in fines of $185m and high-profile resignations, including its CEO.

Investors had a keen eye on the political reaction to the scandal. Earlier this year, Wells Fargo was back in the headlines when news broke that district attorneys in two US states would be investigating the company, this time for improper mortgage lending allegations. Wells Fargo faced another huge hit – $1.2bn in settlement claims.

Naturally, clients wanted to understand the nature of these investigations, the possible outcomes and whether investors who held stock in the bank should be worried.

Experts, Experts, Experts

In our review, we found that experts in banking regulation and compliance were able to shine a light on what was at stake. Specifically, a former compliance and risk manager at Wells Fargo Bank was able to discuss the implications of the scandal – both for the bank itself and the industry as a whole – and a former senior vice president at PNC Bank was able to draw one. experience and offer insights as to how reputation-damaging scandals like these impacts client retention, even within the retail banking sector where customers tend to be very loyal. 

What would a computer programme need to even get close to that kind of insight? Even if we assume that we might one day be able to quantify reputation and client ‘stickiness’, the programme would need reams of similar cases to build up a picture that can be applied to new cases. How often do scandals like this erupt? It would take decades to craft a system that even remotely matches up to human experience. And that’s without even starting on those assumptions.

Ultimately, there’s a reason most investors still trust humans above all else, and it’s not just because they (the investors) are slow to change. They already know and trust the value they get from other people. A CFA Institute survey last year found that, over a three-year period, 69% of investors in the UK prefer the help of an investment professional over the latest technologies and tools. This figure stands at 73% in the US and as high as 81% in Canada.

Still Our World

There’s little reason to think reliance on human beings will change much beyond those three years. McKinsey recently tried to weigh the potential for automation in a wide range of human activities, covering some 800 different occupations in the US. The results around data collection and processing were predictable, but they found no evidence that use of expertise and management of others could feasibly be automated.

Image result for mckinsey automation human
Source: McKinsey

 

The findings reflect, more or less, what is happening in the investing world. Quant approaches to investing and research stand to be dominated by machines as technology advances. However, people will still rule when it comes to using expertise.

Ideas like Big Data are immensely powerful and certainly have lots of potential. But, if we are just riding the crest of another boom in unrealistic tech expectations, then it’s important to think critically about what one is committing to when trusting investments to data and algorithms. Even if we’re not in a bubble that is ready to burst, there is no need to risk all of our chips on the digital world. There will always be a place for human insight, hand-in-hand with whatever the future brings.