An Examination of Job Automation: The Future of the Financial Analyst
I’ve discussed job automation several times in the past, from looking at Bill Gates’ suggestion that we should tax robots (here) to questioning whether Trump’s attempts to bring US corporation outsourced jobs back to American shores addresses the real issue (here).
I think it’s fair to say that as a species, we are hitting a period of exponential technological capabilities which are endangering the livelihoods of people like never before. Human ingenuity has always managed to keep us ahead of the curve and made for additional work as technology replaces the grunt work of manual labour that existed in many industries.
The difficulty is that as the technologies which replace us evolve, the jobs that they do become ever more advanced. At some point, the question has to arise as to whether we have effectively created our own unemployment and given that human intelligence likely approximates some form of Gaussian (or normal) distribution it’s quite possible that over time we will basically put the species out of a job except for all but a very few at the extreme end of the bell curve.
From the Spear to the AI Finance Analyst
As I’m sure we’re all aware, humans are such a successful species because we had the ability to out evolve our other planetary counterparts not in the traditional sense, but in the brain department. Physical advantages are yesterday’s news. We’re not the fastest runners, the best swimmers or the highest fliers. We don’t have the best eyesight, keenest sense of smell or hardiest bodies to survive more extreme conditions than other life forms on Earth. But we don’t need any of that rubbish, because we have the ability to mentally abstract ourselves out of given problem scenarios and construct tools and technologies to get our frail bodies through the situation in question.
In the early days the focus of course was on the basics. Weapons to help us hunt animals for food, build shelter from weather extremes and other things to make the necessities of survival less of an ordeal. As we became experts in those things, we moved on to the next difficulties that we faced until we are now at the stage where the driving goal is increasingly becoming less about the convenience of the species and more the profit motive. Under the proven human construct of capitalism, this is of course an expected outcome to drive the profit motive.
Human technological advancement has taken us on a journey from the humble spear through the washing machine, ATMs and automated store checkouts to the automated contract lawyer, algorithmic trading engine and now, the AI equities analyst. As someone who has spent a reasonable amount of my career designing finance technology to automate jobs out of existence, it’s an interesting scenario to look at.
AIERA: The Wells Fargo Equities Robo-Analyst
A team of equity analysts at Wells Fargo (NYSE:WFC) recently unveiled the Artificially Intelligent Equity Research Analyst (AIERA) to clients. At the moment, it’s more a case study with the goal of getting AIERA (dubbed a “she”) to help them sift through the vast amounts of data that exist in today’s world and help the team make better analyses. This of course is an interim step to the longer term goal of replacement, both for the purposes of removing the human possibility of having a “bad day” to the obvious efficiency gains and cost savings associated with automating jobs out of existence.
Although AIERA isn’t a stock picker in the same vein as an actual analyst, the research team that designed and built “her” pointed out that despite only being 6 months old, validity testing continues to indicate above average results. Despite this, it may be a touch early to put faith in the contrarian call AIERA made last week to sell Facebook (NASDAQ:FB) and Google (NASDAQ:GOOGL), putting it in direct opposition with its human masters at Wells Fargo (not to mention Credit Suisse (VTX:CSGN) who this week upgraded both names to new price targets at Street highs).
But of course the battle between humans and technologies is being waged on many fronts. Whether you put your pension fund into mostly tracker funds or actively managed portfolios run by a human. It does feel like it’s mostly a matter of time before the latest battleground is ceded to technology.
Where Do We Go From Here?
To paraphrase Heraclitus, “The one constant is change”. People of both left and right think that at some point, a new economic model is inevitable as jobs continue to get automated out of existence. Calls for guaranteed minimum income are of course hugely premature, as well as being limited by conventional thinking and the current model. Global GDP per capita stood at a paltry $15,800 in 2015, much of that still driven by human workers despite the leaps in automation that have occurred globally.
The automation of data analytics in many ways is a foregone conclusion. We live in a big data world and there is simply too much volume of data for humans to be able to encompass it all and make appropriate calls. Additionally, the human mind is subject to significant bias and therein lies much of the problem. The brain is a great machine for context switching between predicate and fuzzy logic but contains significant biases associated with everything from life events to conventional wisdom. That makes for a very haphazard decision engine on a large enough scale.
Many argue that AI and machines will find it difficult to make the so called “intuitive leap” that the human mind is capable of but these arguments misunderstand the basic nature of the problem. Computers have been making the news since Deep Blue beat Kasparov at Chess, but these are not intuitive leaps, these are brute force scenario calculations. Until AlphaGo came along, it had been surmised that computers wouldn’t be capable of the pattern recognition and learning required to succeed against humans in the significantly more complex game of Go. Yet that was another hurdle technology overcame with relative ease. The problem of course, isn't whether computers are better at intuition than humans, but whether they can sufficiently exceed human levels of performance by approximating it.
Simply put, the scenarios of human superiority over technology are still huge, but dwindling daily and at an increasingly rapid pace. Is AIERA ready to replace humans? No, there are a huge amount of both technical and regulatory obstacles before that situation comes about, but increasingly we are all working towards our own professional obsolescence and we’d do well to keep that in mind.