COGITO ERGO SUM - Will computers make our jobs redundant?

Descartes’ phrase “I think therefore I am” implies that our ability to think and rationalise is the only undeniable proof of existence. By thinking, I think he meant self-awareness. It is enjoyable, if disconcerting, to speculate about the implications of the development of artificial intelligence (AI) systems developing self awareness through logic.  While computers have yet to pass the Turing test, (Alan Turing’s 1950 proposed  test of a machine's ability to exhibit intelligent behaviour indistinguishable from that of a human), new developments in AI are bringing us ever closer to that point.

There has been some fairly disconcerting commentary on the possible impact of AI, but if one puts the potential destruction of humanity aside (a big if), we should be considering how to deploy such developing technology and (selfishly) the impact it could have on professions.

Advances in the ability of software programmes to include sophisticated decision trees has already driven outcomes in certain areas which are not only indistinguishable from those generated by a human, but in some cases preferable, as they can be faster and more consistent.  However such rules-based approaches are being superseded by machine and deep learning techniques.  Artificial neural networks enable programmes to demonstrate the ability to “learn” and become more efficient and accurate in undertaking tasks over time without the input of a programmer.  The shortcoming of rules-based systems is that we are attempting to model the world in its almost infinite complexity.  Thus rules-based systems are fundamentally constrained by programmer input while machine learning theoretically sheds these shackles.  In a machine learning environment, what is being attempted is modelling the brain (to observe the world).  The potential for AI to take on “thinking” tasks from humans is clear. 

Machine learning techniques are taking off now as a result of two key factors.  First humanity is generating and transmitting simply huge amounts of data nowadays.  These data sets are a pre-requisite for training a neural network to learn.  Second, one requires enormous computing power to process that data in a timely fashion and it turns out that Graphic Processing Units (initially developed to enhance 3D gaming) are ideally suited to the way the data needs to be handled and are improving speed exponentially.  Without these, machine learning exists only in theory. 
 
AI is here to stay and may be to the professions in the current era what the introduction of robots was to the manufacturing industry in the latter part of the 20th century. In wealth management, the phrase “robo-advice” explicitly alludes to this comparison. 
 
The ability to access and process huge amounts of historical and contemporary data, if combined with a sufficiently robust AI programme, will certainly impact some facets of professional life currently delivered by us evolved apes. However, as was the case with robotics in the mid to late 20th century industrial evolution, we are far from displacing all human input.
 
At the moment, and to try to ensure positive client outcomes, professional bodies, in tandem with regulators, ensure a scarcity of those qualified to access some areas of information and knowledge. This scarcity puts a premium on the members of those bodies particularly in the absence of an efficient channel to interpret the raw information for clients. So the gatekeepers, to what in other contexts might be “commodity” knowledge, can earn outsized rewards and keep costs high. AI potentially provides the channel through which technical information can be interpreted “safely” and efficiently for clients, without the direct need for expensive human intermediation.  Outside the simpler technical outputs, humans will still be required to deliver complex solutions, drive innovation and tailor general tenets to individual circumstance. 
 
AI developments should therefore democratise much hitherto protected knowledge, and as supply of knowledge increases its cost will fall. This could be construed as an existential challenge to the professions. However, viewed as a tool to improve service and efficiency AI is more of an opportunity than a fundamental threat. It can be developed specifically to answer unmet needs, as well as to radically improve the efficiency of current services. 
 
AI will be, and is, able to process large amounts of data far better than humans. It takes emotion out of decisions and is therefore, subject to programming parameters, consistent. However, as currently configured, AI will fall short in reacting to new information, proposing novel solutions and be unable to deliver the empathy and understanding that some complex client interactions require. 
 
In the wealth and investment management industry, we are already seeing the inroads of robo-advice into financial planning and to a smaller extent, investment management.  My contention is that we should embrace rather than fear it. Our profession needs to understand what AI can deliver, where it can add value and use it as something with which we can deliver better outcomes to clients at lower cost. 

There is a core of essentially commoditised work and service that is not available to some clients on grounds of cost. AI can address this directly and could be a profound positive force. It will disrupt much of the commodity side of the industry but equally create a new market for clients who today are uneconomic.

Consider financial planning.  The structure of the profession and its current regulation implies a reasonable level of fixed cost.  This has put professional financial planning beyond the reach of many less wealthy individuals, Even potential clients at the start of their wealth accumulation can find it hard to access advice.  This is a  failing of the industry and as pension auto-enrolment rolls on, access to reliable and cheap advice will only become more important. AI is perfectly placed to provide basic advice for individuals with simpler needs or reduced options.  Equally, by removing repetitive or standard tasks from humans it has a place in reducing the cost of initial and ongoing planning. 
 
Using AI soundly means industry professionals can focus on adding value to clients, particularly during those significant life changing events where all the powers of empathy, understanding and nuance must be brought to bear – and which AI cannot deliver. Industry winners will be those that can combine the efficiencies of AI with personal touch, trust and human insight to add real value to clients. Pure models either way will struggle. 
 
The adoption of AI programmes into financial services brings interesting issues similar to those in other spheres. For example, a sticking point in in the adoption of so-called driverless cars is apportioning liability in the event of accident. If someone is killed in an accident who is liable? Is it the company that deployed the technology, the person who instructed the journey (the passenger), the programmer who decided where the vehicle’s safety priorities lay or the constructor of the ethical framework within which the programmer decided on priorities?  Without clarity, no-one will offer insurance policies.

In wealth management, where advice has been given through an AI system, liability may need to be apportioned in the event of error, market action (given the regulator’s outcomes based focus) or even non timely updates of tax, regulatory or legal changes.  Where that liability sits will need to be understood or insurers will begin to withdraw professional indemnity policies. 
 
AI has the potential to markedly boost the efficiency of current business done, commoditising and automating much of what is now a costly and labour intensive process.  From standard advice, through routine asset allocation and re-balancing of portfolios AI can free professionals to do more of what clients pay them for. 

New technology is always a threat to those that mis-understand or fail to adapt to accommodate it.  My contention is that artificial intelligence techniques offer the financial services industry and its clients huge benefits.  In as far as it promotes efficiency, costs and fees should fall with no impact on the technical delivery of services.  A corollary of this will be the extension of advice to a new cohort of clients who are currently priced out of the market.  We should applaud this. 

As with most innovation it will take time for the optimal business models incorporating AI to evolve (think of the initial attempts of retailers to grasp the internet opportunity),  A big challenge is going to be understanding pricing dynamics, ensuring that commoditising one part of the proposition does not undermine pricing in other parts where professional input remains vital and where higher fees are appropriate and legitimate.

It is early in the development of AI tools in the financial services industry, but businesses should already be taking its adoption seriously and planning their business models in light of it. 

Companies with confidence in their ability to add value will embrace this coming opportunity and benefit as a result. Companies that do not will be competed away.

Andrew Herberts

Head of Private Client Investment Management (UK)

 

Article first appeared in:

Expert Investor Europe (in print on the 1st January 2018)

 

Clients are advised that the value of all investments can go up as well as down. Any past performance or yields quoted should not be considered reliable indicators of future returns. Opinions, interpretations and conclusions expressed in this document represent our judgement as of this date and are subject to change. Furthermore, the content is not intended to be relied upon as a forecast, research or investment advice, and is not a recommendation, offer or a solicitation to buy or sell any securities or to adopt any investment strategy.

Thomas Miller Investment is the trading name of the businesses in the Thomas Miller Investment Group. Thomas Miller Wealth Management Limited is authorised and regulated by the Financial Conduct Authority (Financial Services Register Number 594155). It is a company registered in England, number 08284862. Thomas Miller Investment Ltd is authorised and regulated by the Financial Conduct Authority (Financial Services Register number 189829). It is a company registered in England, number 2187502. The registered office for both companies is 90 Fenchurch Street, London EC3M 4ST. Thomas Miller Investment (Isle of Man) Limited is licensed by the Isle of Man Financial Services Authority. It is a company registered in the Isle of Man, number 48181C. The registered office is Level 2, Samuel Harris House 5-11 St Georges Street, Douglas, Isle of Man, IM1 1AJ. Thomas Miller Investment is a registered business name of Thomas Miller Investment (Isle of Man) Limited. Telephone calls may be recorded.