Financial Services Must Get More Out of Their Data - AI Holds The Key

Ciarán Daly

October 27, 2017

9 Min Read

With estimates showing that algorithmic trading systems handle 75% of the volume of global trades worldwide, financial services are feeling the impact of AI investment more than most. There's a number of reasons for this, not least because finance work is incredibly data-heavy.

"Yesterday, I had an interesting conversation," offers Dr. Christian Westermann, Data & Analytics Partner with PwC Switzerland. He is better-placed than many to explain why AI is already such an area of interest within financial services. Formerly a space scientist, Westermann oversaw the early development of  data analytics as a topic of interest at the firm from 2009 onwards. As the data and analytics leader, he's ultimately responsible for all artificial intelligence within the Swiss PwC practice - which itself specialises in banking, insurance, and pharma life sciences.

"I was presenting, speaking about AI and so on, and one audience member - he was from a bank - said to me, 'look! we are basically a data bank!'. What he wanted to say - and I can truly support it - is that, at the end of the day, a bank is very virtual. It only consists of data," explains Westermann. "There's nothing tangible there that you can really touch or feel, so everything, even the products offered, the investments, the interest paid, the loans given... 99% of it is truly virtual. It's true data. Financial services consist of data, and should be the ones who are highly interested in doing more with their data."

'Tremendous Potential' For AI In Financial Services

Westermann believes this is just a precondition of why there is so much interest in AI from the sector. He believes that financial services firms have what he calls "tremendous potential" to increase their productivity. "Banks are quite complex. Their infrastructure is quite complex. The compliance processes are complex. The risk processes are complex. These days, with all the regulation, banking has become quite expensive." Citing the popular FS adage that, behind every one financial relationship, you have three people from compliance who need to protect the different aspects of what the relationship is actually accomplishing, Westermann believes that the sector's demand to become more automated and linear is self-explanatory. AI and machine learning offer a means of increasing productivity and streamlining those aspects of the sector.

"The third factor that puts AI and machine learning in focus within financial services industries is consumer demand," Westermann argues. The rapidly changing landscape of financial services - largely thanks to disruption from borderless fintech companies - means that banks no longer have the automatic right to 'win'. In other words, the landscape is shifting - and fast. "This is the time of the sharing economy, and there are many reasons why banks aren't needed in these times. They are facing competition from everywhere. If we look at the key driver of AI - changing consumer demand - banks must immediately do something. They must get more out of their data - otherwise, they will face huge challenges from new players in the market, primarily fintech startups but also other companies."

Data Provides Three Key Challenges For Financial Services

Westermann argues that one of the greatest challenges for financial services firms is that of leveraging their data: primarily, developing good data quality and governance. It is these factors, he argues, which are fundamental to the use of AI. "To a large extent, AI doesn't help you to achieve better data governance or quality, but if you don't have good data quality governance, then AI can never reach its potential," says Westermann.

The next key challenge, he explains, is bridging the gap between divisions within banks and insurance firms. "Whether it's investment banking, retail banking, or wealth management, banks need to work across their divisions."

"The third, and perhaps one of the most important challenges, is everything that comes from the regulator," Westermann says. "Often, what comes from regulators has a perspective which is going in the opposite direction from those in financial services. The need for data-driven use cases faces all the limitations given by the regulator."

In many ways, he is correct - the kinds of concerns central to regulators are far behind the kind of concerns AI is practically going to raise. Take the example of GDPR, i.e. the EU's General Data Protecting Regulation. In the near future, this will offer clients a 'right to explanation' about algorithmic decisions made about them. But, Westermann warns, the very nature of AI provides a number of hurdles to ensuring these regulations are followed. "If we think about AI, we cannot explain exactly what neural networks are doing, because the algorithms are big complex formulas and their decision criteria is not really clear," he explains. "On the other hand, banks need to comply with GDPR."

It's a dilemma for many within the sector, and it's why PwC have developed their Responsible AI framework. It provides a number of foundational principles for AI design and implementation within organisations, focused on transparency, user experience, regulatory compliance, and more. Westermann argues that the framework will "directly help financial services organisations navigate through those regulations so AI can really meet its potential." Furthermore, AI itself can be used to better match external demands coming from regulators with internal policies and procedures.

AI Is No Fix-All - But Its Potential Must Be Unleashed

It's important to note that AI is not a fix-all for the obstacles facing financial services. "To a large extent, AI cannot overcome these fundamental challenges within financial services," admits Westermann. "But there are challenges that FS firms can overcome to ensure that AI really unleashes its potential. Data strategy really needs to be in line and in sync with business strategy. Only if it follows the business strategy can data strategy make sense."

No matter which objectives underpin the business strategy - to be more efficient, to provide more individual levels of service - data strategy must be aligned to those objectives. "Only then will the potential of AI become clear."

So what might this look like? How can financial services successfully implement an AI-ready data strategy? Westermann argues that, firstly, companies need to prioritise the development of a trusted data foundation, which needs to account for data governance. "For instance, it can provide a framework in which AI applications can be developed in a safe way, providing go-to politics which allow the exchange of data and make it available across divisions." In other words, it provides the rules for how divisions should cooperate when exchanging data. "Through this trusted data foundation, the goal should really be to provide an environment that in the later stage, the use cases that really should drive the bank can be accelerated. That is the first and very important element of the trusted data foundation."

A successful AI-ready data strategy requires more than just good data governance. Westermann believes that banks must be ready for big organisational shifts in order to ensure data and business strategy are firmly aligned. "The bank needs to have the right organisation. This can be achieved through upskilling people, having right skills onboard, skillswapping, maybe even through the help of externals over time," he explains. It also means having the right people in the organisation that possess what he calls a "data and innovation-driven mindset".

Finally, financial services organisations need to provide the right infrastructure, tools, and applications for AI development - what Westermann calls 'ideation areas'. This is where personnel are able to develop and test AI applications on an experimental level, providing the relevant stakeholders access to areas where, in Westermann's words, "they can play around with new apps, test new products, and make them production ready. In other words, providing the reference architecture - the data environment - in which AI use cases can really be developed."

"These experimental areas are very useful because, in a very short time period, you can start to validate the applications you are building. You're able to test your minimum viable products with the true stakeholders. That's a little bit different to traditional IT, where things are developed for a long time in a development environment and can only be used after some money has gone into them," he explains. "What is really anticipated is that, in a much more efficient way, the user behind it has the chance to play around with the new technologies - and new ideas - that will come up. With that, developers and innovators get much faster feedback, whether the application they are driving is going in the right direction, or whether there is a way, after a soft failure, an app is stopped because stakeholders don't value it."

Benefits Of AI In Financial Services

This year, PwC's Global Artificial Intelligence Study provided a number of exceptional insights, asking clients across many different territories where they believe the benefits of AI will fall over the next few years. "What do you really solve with AI?" Westermann asks. "Two driving factors came out that will really influence the industry over the next fifteen years. On the one hand, there's productivity gains - AI will be the technology that supports productivity gains. The second element that drives AI will be consumer demand. We as consumers are already using AI. Everything that comes back through Google and Siri, even the smart tools in our cars... Already, our demands as consumers show how we are dependent upon AI." PwC see this as a large influencing factor driving demand over the next few years.

"Banks and financial services companies, they will be driven by the need for productivity gains and this change in consumer demand," Westermann says. "This relates back to data strategy, which should have these two goals in focus - because from there, the demand will emerge."

Dr. Christian Westermann is Partner of Data and Analytics at PwC Switzerland. He will be speaking at next month's AI Finance Summit in Zurich.

Christian_Westermann_0027-300x451.jpg

Christian is a Partner in PwC Digital Services. He has a PhD in Physics/Space Research from the University of Bern, where he was part of a team that built satellite-based instruments for ESA and NASA space missions. In 2005 Christian also completed an Executive Bachelor in Business Administration.

Christian has fifteen years of experience in data analytics and IT management. In his career he has managed a number of large-scaled projects in the Financial Services Industry.

Christian is leading the Data & Analytics practice for PwC Switzerland – a team of 80 data and modelling specialists based in Zurich, with hands-on experience in Machine Learning, Deep Learning, Robotics, Natural Language Processing, Process Intelligence, IoT, Customer Analytics and Simulation & Modelling.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like