It’s encouraging that Britain’s banking watchdog, the Financial Conduct Authority (FCA), is backing more aggressive deployment of the weapons of artificial intelligence (AI) and machine learning (ML) in the fight against crime. However, this poses a dilemma for the main players in the financial services industry.

While, the FCA has been demanding the utmost transparency from banks for years, most recently it argued that AI and ML algorithms can do so much more to help protect clients, with all the enthusiasm that a foreign exchange dealer might pursue profits.

It is not beyond the wit and imagination of banks to use automated decision making. Indeed, they already do this for many of their trading systems. However, there is an argument that the interests of customers are at risk from feeble efforts at automation.

It is indeed positive to hear that the FCA seems to have stiffened its sinews and summoned the blood to get tough in the war against cybercrime. The hackers, crackers and blackmailers of the crime industry are at the cutting edge of artificial intelligence tools and there is a danger that a sophistication gap will open up and the criminals become light years ahead of the banks.

Some might attribute this to a lack of urgency in the banks’ response. When it comes to exploiting the wondrous potential of artificial intelligence and machine learning, it might be the case that safety is not the priority. There may be more exciting, money-spinning options available for using this powerful technology.

That would certainly be one explanation as to why many banks are still taking manual approaches to crime busting. However, there is some logic and merit to taking a manual approach. After all, the defence of a company against intruders is a constant work in progress and banks are on a permanent learning curve. While AI and ML are undoubtedly good, the breadth and variety of learning potential for humans far exceeds that of machines.

Consider the types of learning techniques that are adopted by humans. Some of us are Sensory learners, other are Intuitive. There is no wrong or right way in the variety of choice we humans have for systems of acquiring knowledge and understanding. We can choose between techniques for visual and verbal learning, active and reflective methods of digesting data, and choose between taking either a sequential or global view of the big picture. Or you can use, as most humans do, a blend of learning techniques.

By contrast, AI is still in the early stages of development. So one can see the case for using humans to investigate sophisticated computer aided financial scams. Which is what seems to have happened in The City of London and Wall Street.

However, there is one area where AI and ML trump manual intervention and that is in the sheer power, speed and durability of the machine. They can relentlessly chase endless leads, while human investigators naturally fatigue and lose accuracy over time. This is why the FCA has recognised AI and ML as crucial weapons against criminals. Not least because criminals themselves are using automation techniques with increasing sophistication.

Just as various types of learning techniques can be blended, it’s high time that we used software to augment human efforts.

The good news is that regulators, such as the UK’s FCA and Singapore’s MAS, are encouraging the development and refinement of AI and ML solutions, with initiatives such as sandboxes and hackathons.

But they will need a good engine to speed up the process of spotting dubious digital transactions. They will need machine learning tools that can instantly scan huge volumes of data, spot subtle patterns and filter transactions. More importantly,  to keep pace these machines must grow stronger with each experience, learning the emerging patterns of fraud so they can stop attacks before they make an impact.

And that is something that companies like TIBCO have spent decades learning how to do: building the platforms and crafting the tools, based on a collective wealth of wisdom that is the product of millions of man hours. Only then, is it possible to see the problem from all angles and on every level from global to sequential. When you’re fighting crime it’s good to arm yourself with every type of weapon - whether it’s sensory or machinery.