In the blink of an AI


Artificial intelligence is already a reality in daily life, and it has a place in financial services, says Matt Davey of Societe Generale Securities Services

What kind of issues do you think the financial services industry should be focusing on, with regards to artificial intelligence?

Artificial intelligence (AI) is a very hot topic at the moment, and there is a lot of talk about it. It encapsulates a lot of the data science and big data initiatives that firms have already been working on, but there is still a question as to what exactly it is. Are people actually using AI, or are they simply using algorithms and process automation? I think there is an interesting debate around that.

Looking at the big picture, banks are all working with legacy systems, and one of the big challenges for them is considering how far they should go in automating and improving those legacy systems, and at what point they should make the jump to a new system. These are complicated legacy infrastructures running large volumes of data, and the big prize is in figuring out how to simplify that and reduce the associated costs.

Another aspect for the large banks is the introduction of voice and image recognition systems using AI. I would question where exactly that sits. I’m not sure that our operational processes lend themselves to AI algorithms for images, and institutional clients don’t necessarily have the same application for voice recognition technologies as retail clients. It will be interesting to see how those tools can be applied to the institutional financial services industry.

Are there any AI applications ready to be applied to the back office?

There are a number of interesting AI applications that could lend themselves to the back office. One is in customer focus—looking across various systems that interface with customers in order to judge their mood. We could potentially predict when we may be likely to lose customers, based on the nature of their interactions, or we could synthesise data to alert people to a trend, which could lead to potentially cross-selling products. Being able to look across a lot of data points could be very valuable for a large bank. Financial crime is another area—pattern detection can be very effective for detecting money laundering and fraud prevention. It might not be obvious to a human being, but using machine learning we could analyse large amounts of data and identify a combination of factors that indicate a higher risk of fraud.

In the retail space, AI has been implemented through call centres and chatbots, for example, and the industry has got used to that to some extent.

Now, that technology is being combined with large data sets and AI in order to create something with a much greater level of intelligence that is far more useful.

Finally, there is the area of regulatory compliance and risk reporting. We have a complicated risk environment and we have seen several years of increasing regulation designed to reduce risk in the industry. That’s likely to be an area where machine learning is used to digest complex regulatory requirements in order to aid compliance.

Do you think there’s a risk of the industry becoming over-reliant on AI technology?

My first car was an Austin Montego, which had a black box engine management system. I was driving in the Netherlands, and the car broke down. I took it to a garage, where the mechanic opened the bonnet, looked at the black box and concluded that he couldn’t do anything for me because he didn’t have a laptop to plug in to run a diagnostic.

That’s a good analogy for the kind of issue we could run into in terms of becoming too reliant on technology. I don’t think we’re at that point yet, but it’s certainly an issue, not just for financial services but more generally as AI becomes more embedded in our lives.

When we look at improving operational efficiency in banks, one of the things we’re working on is robotic process agents (RPA). You can create a robotic agent to manage a process, and effectively lay this on top of existing IT systems in order to make them more efficient and to increase their longevity.

There is a double edge here, though, as it also fixes that IT system in its current state, making it harder to change the processes later.

Will the industry still need human interaction, no matter what happens?

This is another philosophical question that can be applied to AI generally. I see it being implemented in a role that supports human decision-making. If you look at the health and medicine industry, for example, it could be incredibly beneficial to have some form of AI to review 70,000 journal articles that reference a particular condition, helping a physician to ultimately make a diagnosis.

Similarly, in chess, we saw brute computing power defeat grand chess masters 20 years ago. But now, if you look at ‘freestyle’ chess, where players can compete in any way or in any combination they like, the most successful teams are those with a combination of human and AI players.

There are some interesting examples of AI supporting humans successfully, but the interesting debate will be around how it’s going to work.

Take up of blockchain technology in financial services has been slow. Is the industry ready for AI as well?

The two are quite different things. There are a few big items to resolve around blockchain—mainly confidentiality and scale—and there are teams working very hard to resolve those. There is a lot of momentum behind blockchain initiatives, but it’s hard to change horses. There are large amounts of payments and assets maintained on existing systems, so you have to be very careful before making any changes to those systems. Switching to blockchain would mean a completely different way of managing transactions, so I wouldn’t expect things to change very quickly. But, that’s not to say it hasn’t got huge potential.

AI is very different. The author William Gibson once said: ‘The future is already here, it’s just not very evenly distributed.’

We already have AI in a lot of aspects of our lives, whether that’s self-driving cars, Siri or prevention of credit card fraud. The challenge is applying it where the business needs it. It can sometimes feel like AI is a solution looking for a problem, and in traditional systems development, the first question is always around the user requirement. What is the user trying to do? There is a danger that, with all the excitement around AI, people are forgetting that.

We need to consider the user requirement and the options that are available. The answer may well be in AI, but it may also, be something much more mundane, such as improving the interface between two systems.
Features
The latest features from Asset Servicing Times
Industry participants discuss how factors, such as financial technology, have changed the role of transfer agents
After many years in the making, MiFID II has now become part of the regulatory furniture in Europe. However, the regulation’s first 30 days revealed lingering concerns around the scope of the changes
Join Our Newsletter

Sign up today and never
miss the latest news or an issue again

Subscribe now
Although MiFID II came into play on 3 January this year, now is not the time to relax as the job is only half done, says Peter Moss of SmartStream
With 2018 being a busy year for regulation in the financial services industry Jon Trinder and Matt Gibbs of Linedata discuss what firms should be working on around MiFID II, as well as the roll out of GDPR in May
With the industry in perpetual change, it’s important for firms to work together to take advantage of big data, according to Roy Kirby of SIX
As technology developments shape the world around them, financial services firms are starting to adapt. This year’s Sibos conference outlined where the industry is settling in, and where there are still milestones to pass
Rocky Martinez considers how AI can help improve post-trade processes, and how SmartStream’s new reconciliations solution is moving a step in the right direction towards helping firms keep costs at sustainable levels
New regulations, new competition and new cost pressures mean custodians and sub-custodians have more balls in the air than ever before
View features section
Country profiles
The latest country profiles from Asset Servicing Times
Luxembourg’s asset servicing industry has blossomed to boast a substantial portion of types of funds in need of back and middle office functions, and now Brexit may offer an opportunity for further growth
In order to stand out from competitors, Malta has resorted to innovative regulation, says to Joseph Camilleri of BOV Fund Services
Securities Lending Times

Visit our sister site
for all the latest securities lending news and analysis

securitieslendingtimes.com
Julien Kasparian, head of Hong Kong at BNP Paribas, explains the financial services industry in Hong Kong and how it competes globally
The Asian market may be improving on the harmonisation front, but the situation is still far from ideal. Experts discuss what there is still left to do
Brazil is hogging the limelight from its South American neighbours. But, although reforms are in full swing, there is still work to be done
No nation is an island, and the Polish CSD has post-trade services to cater to all of Central and Eastern Europe, says KDPW’s Iwona Sroka
In a region as geographically, culturally and economically diverse as Asia, funds passports have a tricky road ahead if they’re to redefine the industry
Amid cross-border restrictions and tightened belts, Luxembourg’s kingdom of real estate investment won’t be crumbling any time soon
View country profiles section
Interviews
The latest interviews from Asset Servicing Times
Chris Meader, founder of the North American Fund Administration Association, discusses his new start-up association, which aims to provide a forum for administrators to understand the risks, concerns and the opportunities the industry is facing
View interviews section