Artificial Intelligence (AI) can do all the things humans do better, faster, and more reliably.
That’s a bold statement.
Because it seems that the next wave of Fintech is aiming to sweep humans out of financial services and who knows out of what else. One reads endless articles imagining dystopian futures where labour, laws and life are dictated by “minds” made of silicon and whose blood, if any, isn’t liquid but an algorithm.
The world is fearful these days. In the midst of the health crisis of 2020, digitalisation has increased dramatically, as have the chances for the kingdom of algorithms to establish itself with little resistance. But is it like that, really?
Although playing with alternative and almost apocalyptic futures is interesting, and even positive, because it pushes us to think about our capacity for response and resilience, we increasingly see the initial fear “AI will replace humans and take their jobs” becoming “AI is a tool to augment human capabilities”. Augmented intelligence, researchers call it.
In the AI race in business, champions want more than implementing AI-powered software to automate cognitive and physical tasks. They need the human touch—and the brain, too.
Before computers, decisions were a result of experience and intuition. But, from ENIAC—the first electronic programmable computer—until the present, the process of answering questions like “risky or safe” or “positive or negative” has greatly changed. That has a lot to do with computers’ ability to process large amounts of data and to narrow down complexity. The resulting options allow the human brain to discern better, and to focus on the aspects of work where intuition may be an advantage.
Thanks to computers, human decision making improves in many ways. We can have control over the final outcome or over certain parts of processes where human thinking can add value. Unquestionably, systems benefit from keeping humans in the loop. Since computers increasingly simulate human thinking skills and are “learning to learn”—that’s arguably the most simplistic but illustrative way to explain what AI does —businesses have to figure out how and where in both their processes and workflows AI enhances efficiency and performance.
Understanding workflows—a series of repeatable tasks or activities in which data is transferred between humans and/or systems—is vital for businesses to truly harness the power of AI. There will be cases when humans will have to take a step aside, and others when their input will be very necessary.
It isn’t a surprise that AI has tempted finance too. It has primarily taken the shape of support systems for investment decisions. Among them, the most conspicuous example we can cite is the robo-advisor. This article wants to get our readers closer to them, but it also offers other examples of AI supporting decision making. Let’s dig in.
Getting to know Robo-advisors
2008 was the year when the first reported robo-advisor, Betterment, was born. Those were turbulent times, much like the ones we’re living in 2020. A microscopic virus has replaced the numerous subprime mortgages irresponsibly granted to people with poor credit scores that ultimately ended up in a global liquidity crisis.
Recalling the context in which robo-advisors appeared isn’t cosmetic. In fact, a growing lack of trust in the financial system set the stage for them to take off or, at least, to find a niche in the financial world. More than a decade later, they are still here, stepping harder and harder.
According to Statista, Assets under management in the robo-advisors segment are expected to show an annual growth rate (CAGR 2020-2024) of 26.0% resulting in a projected total amount of US$2,487 billion by 2024.
Good example of decision support systems, robo-advisors are software running economic models in the background. More and more, those economic models are augmented by trained AI systems. Simply put, they are automated investment platforms that provide online portfolio management with minimal human intervention, using trading algorithms. Users go through a self-assessment process so they can make decisions from various investment alternatives.
Robo-advisors are trained with thousands of priorly given advice and, thanks to AI technology, can come up with options adjusted to a user’s goals. They are the “outsourcees” of investment advice at retail level.
Robo-advisors democratise investment advice
According to this PwC study, prominent features distinguish a robo-advisor from a traditional financial advisor. We want to highlight these four:
1) Robo advisors democratise the advisory service. Robo-advisors allow users to invest even with low amounts, expanding the client base and favouring the inclusion of millennials and the Generation Z.
2) They are pretty transparent when it comes to management fees and the way to invoice costs is flexible. Robo-advisors, in fact, commonly offer a flat rate, while the portfolios managed by traditional advisors generate commissions for each transaction, management costs and administration of the securities.
3) They usually offer a remarkable user experience. From effortless access via smartphone apps or compelling websites to easy-to-understand investment products, robo-advisory platforms put users at the center stage.
4) They are educational, explaining contracts and legal aspects linked to the investment in simple terms regardless of the users’ financial background.
Advising the advisor
One may wonder if traditional financial advisors take advantage of the power of robo-advisors too. Well, robo-advisors are primarily meant to be used by end users, let’s say you and us.
Market analysis tools, on the other hand, are mostly used by financial advisors. They give early signals of trends at different scales, analysing information in many different languages. As a result, unmanageable volumes of data are transformed into insights that are digestible for the human brain (and eye), something easy and quick to understand, interpret and, consequently, give advice upon.
Some robo-advisory platforms are blended, i.e. they combine the power of algorithms and human insight. While, in most cases, advising does not require the intervention of humans, there are robo advisors whose role is to propose a list of the best investment alternatives that a traditional financial advisor will analyse afterwards.
Let’s talk about biases and robo-advisors
Not even gigantic amounts of data could shield AI and any technology powered by it from human biases affecting decisions and judgements. Robo advisors are, unavoidably, trained with data that bring with them cognitive biases of different nature.
However, even with this downside, AI-powered decision support systems greatly improve the quality of decisions, augmenting and extending human capabilities. For AI developers, understanding where data comes from, how it has been curated, and how it will be used isn’t only a technical responsibility but also ethical. Researchers are actively working on methods to detect and mitigate biases in AI systems – something that we will hopefully see more frequently in the market in the near future.
Making robo-advisory and other AI-powered systems as unbiased as possible hasn’t quite been achieved yet, so now is not the time for complacency. For instance, in a 2020’s article of the Journal of Behavioral and Experimental Finance article, authors conclude that, for the reality of a developing country such as India, “robo-advisory platforms are not yet comprehensively self-sufficient to accurately perform risk analysis for retail investors”.
On the other hand, when it comes to user adoption, one may think that young, tech-savvy individuals are predominantly robo-advisors users. But, once again, our brains’ biases are playing their cards.
An interesting conclusion of the PwC report mentioned above is that user characteristics—e.g., gender and age—don’t appear to be significant in the attitude formation process towards the use of robo-advisors. On the contrary, variables such as trust and risk are more relevant for adoption. The former is positively correlated with favourable attitude formation; the latter, with a negative one.
Other examples or AI-powered decision support systems
Apart from finance, other industries are tapping into the advantages of decision making assisted by AI. For instance healthcare, marketing, entertainment and communication, e-commerce, command and control, and cybersecurity are already using the also called intelligent decision support systems (IDSS). Here are some examples:
- AI-powered image processing software helps pre-select images so radiologists make faster and better decisions in cancer detection.
- Predictive maintenance. Factories have implemented AI technology in the lines of production to detect potential dysfunctioning and supplies shortages.
- Semi-autonomous cars use image processing to understand, for instance, warning signals and help drivers make decisions. In this case, image processing uses deep learning, a type of machine learning.
- Weather forecasting, improving the chances to fight disasters and, therefore, put in place more effective recovery plans. This is bridging the gap between data scientists and climate scientists.
- In marketing, AI assists marketers in modeling buyer personas by analysing, for instance, user behaviour and how users interact across different brands’ touchpoints.
- Content recommendations on streaming services, e-commerce sites and on social media for instance. Indeed, AI, based on consumer insights, purchasing or consumption patterns, gives tailored suggestions.
Once, a colleague of ours referred to robo-advisors as the centaurs of retail investors, a being halfway between robot and person. That immediately prompted our minds to think about what they would look like.
However, in reality, they are simply algorithms and, although they may not be so visually interesting, they are already very useful in improving decision-making in business workflows and other processes.
This isn’t an endgame for humans as data processors, not at all. Business decisions requiring more than just structured data are plentiful and that’s when the beauty of the human mind comes in. For instance, when discussing and defining the organisation’s mission and vision, company strategies, corporate values, etc, to name a few where our background, history, memories and life experiences are invaluable. But maybe AI can help us in finding inspiration from the millions of examples on company strategies it has analysed and is presenting to us in a digestible way.
AI and humans, in coordination, can make better decisions than each of them alone.
What we think
The notion of quick and large-scale replacement of expert workers by AI has been just around the corner for several years now. While AI has proven a powerful tool for many tasks, humans excel in many others. Companies are increasingly aware of this fact, so they are redesigning their business processes to have their experts and customers supported by AI in a targeted way. The combination of AI analysing large amounts of data to support humans taking more informed and creative decisions is a powerful one, but it is necessary to integrate this into your organisation in a responsible way.