Over the last couple of years, there has been a buzz of activity around artificial intelligence (AI).
Last year, Facebook founder Mark Zuckerberg used AI to create a personal home assistant. A lot of work still needs to be done, but at least Mr Zuckerberg's considerable financial backing means that there is serious work behind it.
Developments in autonomous driving technology have seen a spate of driverless cars not only in the United States, but also in Singapore. You might have seen signs about driverless cars in the one-north/Biopolis area if you have visited the place recently.
An interesting point about driving is that certain rules sometimes need to be "broken" in order to drive effectively. Please do not misunderstand me. I am not condoning illegal driving here.
Let us assume that a car has broken down in front of you and requires you to overtake, crossing a double white line. The AI controlling the car knows that it is not supposed to cross a double white line yet if it does not do so, it cannot move, so a decision needs to be made here.
Of course a poorly programmed driverless car could wait until the broken-down car is towed away, but that could take hours and I doubt that would be considered an "intelligent" decision.
This decision-making factor is just the tip of the iceberg in demonstrating the difficulty of programming a fully autonomous driverless car. Other factors include the weather - think heavy torrential rain interfering with car sensors - and probably the biggest unpredictable factor - other human drivers - which makes fully autonomous driving incredibly complex.
Besides driverless cars, we have seen the emergence of robo-advisers in the last few months in Singapore.
A robo-adviser is basically a program that helps clients decide on an investment portfolio. There is usually no human intervention and portfolios are decided through mathematical rules or algorithms based on information input by the client.
The execution of the investment portfolio is typically done through low-cost exchange-traded funds (ETFs) and the initial benefits that have been listed include cost savings, 24-hour access and low minimum balance requirements.
Many of these benefits are the result of the removal of the human element - the cost savings related to the absence of the financial adviser, the 24-hour access versus the working hours of their human counterparts, and the minimum balance requirements for an investment adviser's services.
A further benefit touted is the removal of a natural human flaw in the advisory process - human emotion - which can result in bias in portfolios.
While it might seem like a very harsh criticism of the human form, this point has been keenly studied and documented in academia on fear and greed. Many investors tend to sell at the low points when they panic and buy after the market has risen to unsustainable levels.
So, what works better? Human adviser or robo-adviser?
A human adviser, despite not being as efficient as a robo-adviser, has many abilities that a robo-adviser is still years away from replicating.
For a start, a human adviser can read other humans in a way that AI cannot. This plays an important role in discovering and understanding nuances in clients' needs. These nuances, at this point in time, are impossible to program into a machine.
Exposure to more sophisticated investments that need to be assessed based on lock-up tenor and clients' comfort with the investment is also still far from being programmable.
That said, from the onset, robo-advisers were not built to handle sophisticated investment portfolios of high-net-worth clients. They were targeted as a low-cost solution so that all clients would have access to sound financial advice.
Yet I believe many of the benefits can make sense to high-net-worth clients. For example, investment bias should be put into check by ensuring client portfolios are run through similar algorithms to check for bias.
Rather than deciding between human and robo, a hybrid model, where human advisers use digital advisory tools to enhance the quality of advice, can be a good compromise.
Either way, technology will no doubt play a big part in how advisers do their work. Tools should be built to ensure that clients' portfolios are properly analysed. Technology should be used to increase service quality, for example, to update clients on a customised and timely basis. Benchmarks will have to be raised as well to ensure competitiveness, the same way in which the asset management industry has had to show their worth against ETFs in the market. And it will be a challenging but ultimately rewarding journey for those that come up tops.
I attended a seminar recently where the speaker, a professor from the Massachusetts Institute of Technology, spoke about how we are still years away from a fully autonomous driving system for the reasons I listed above. They include real-world decision management, ability of the sensors in different weather conditions and unpredictable human drivers, just to name a few.
However, partial "AI" systems in driving have been around for decades. The anti-lock braking system (ABS) that is in every car today takes control of your braking to prevent skidding in wet weather. This is an example of a partial AI system. It assists when needed but the human is still in control. In my mind, this partial/hybrid model makes good sense even in the wealth advisory field.
There is still much information that a human can bring to the table and coupling this with the efficiency of technology can only result in better quality advice to clients.
Pre-populated model portfolios against client risk classes - which is what robo-advisers deliver - can be the starting point, but the ability to customise the portfolio to the client's needs, either by fine-tuning the risk of the portfolio, or introducing specialised product features, or other non-traditional options, will better meet the needs of the growingly sophisticated investor base.
• The writer is Standard Chartered Bank Singapore's head of investment advisory.