In the mainstream media artificial intelligence (AI) is often viewed as a mythical phenomenon that will solve future problems and that is permitting new companies to replace whatever firms existed before. As we all know, those images fall far from the technologies commonly referred to under the umbrella of AI today. Let’s take a look at what the current capabilities permit us to do in the financial sector.
Human-in-the-loop AI becomes part of the existing workflows
Starting by the nature of AI solutions, the most impactful AI applications are built somewhere around the collaboration of humans and machines. As such, they are not tied to the “born digital” companies only. Any company can create human-in-the-loop AI systems, where the purpose is to elevate the efficiency of an employee into a new level with help of a machine. The machine is able to process much more data in a matter of seconds, and then curate it to the human expert in a more consumable format. AI solutions that are built with humans at their core focus on keeping humans “in the loop”, thus providing them the benefit of using AI tools that will make them more efficient in their work.
Take the example of an investment bank. Today, the decisions for making investments are largely based on the investor going through a vast amount of texts required by law. These long text files, documents and contracts are investigated in order to find facts that might encourage or discourage the investment. When using an AI system for investing, the investor would get pre-processed documents that have the risk factors and recommendations already pointed out from the investment materials. Previously we’ve considered using AI in impact investing.
Virtuous learning loop: risk assessment improves overtime
Any financial institution is built around considering risk. The AI system can include a learning loop, where the investor teaches the machine to recognise pieces in the text that affect directly the investment decision. This way the machine learns to point out better recommendations and improve the investment risk assessment. The investor can focus on drawing conclusions out of the materials presented and pre-processed by the AI system. The employees can also on further educate the machine to become even better at its job. This way the learning loop improves both parties, the human and the machine, to become better at what they are best at.
Accounting is another area where everyday processes can be improved significantly when using machine learning and artificial intelligence. Accounting in large corporates is a massive, complex system with many different kinds of sources and data points to address. Natural Language Processing (NLP) helps the machine to interpret invoices and cost center reports, and make recommendations for finding the correct category for them. When successful, the machine reinforces the positive learning loop, and when not, it adjusts the suggestion accordingly.
To succeed AI needs to be applied on achievable, measurable and impactful opportunities
When talking about artificial intelligence, it is important to see the machine as part of a natural development in us humans leveraging technology in our work. The AI systems should not be taken as something that would replace humans, as that is not realistic. Human can be replaced by sophisticated algorithms in only a fraction of cases where huge data amounts are being analysed as part of a process. It is more important to improve an existing workflow by having AI systems assist employees in being more efficient. Especially in finance, we will always want to work with people for certain things, such as in managing our private wealth.
Finding the best opportunities where to leverage AI systems is a challenge that not many companies can answer themselves. This can lead to trying to use AI in applications, where it is not at its best, and trying to build too fancy, but unnecessary systems, that present a risk for the company. Not only does the experimentation culture suffer, but also the company might lose valuable time and money. Therefore when considering AI opportunities, it is important to take time to find those opportunities that are achievable, clearly measurable and that have a direct impact on improving competitiveness.