Q&A: How AI will transform the financial services industry

What’s unique about AI and ML is that everyone can see how these will change their everyday lives – and that’s very exciting. It’s a conversation topic stretching from the dining room to the boardroom and the highest levels of government.

Generative AI has shown its magical capabilities immediately and spark an innovation and business transformation era on a scale we haven’t seen in years.

Gil is spot-on about the GenAI aspect. What’s really struck me is its accessibility – the fact that even people with a non-technical background can already understand and use it.

I think we’re all familiar with the web search world, where the results are a bit all over the place. But the results are absolutely in context with GenAI. Its responses resonate pretty much immediately and can be fine-tuned, which is like nothing we’ve seen before.

The biggest opportunities exist where the technology can make the day-to-day lives of customers and colleagues easier. FS has traditionally been tagged as a transactional world in which firms have struggled with personalisation, because their data is very fragmented. But we’re moving into a more conversational world with GenAI-enabled banking, where customers can obtain more concise answers and clearer directions.

AI/ML is also being deployed in areas such as fraud detection and ‘know your customer’ processes. This can have a positive impact by streamlining the overall experience of customers while also keeping them safe.

Deutsche Bank aspires to be the leading global responsible AI bank. We are channelling the energy and excitement around AI/ML to drive a responsible AI business and innovation transformation of the bank. That means taking a phased approach and applying GenAI in controlled, well-defined, and non-client facing use cases to establish confidence internally.

For example, we’re experimenting with chatbots across internal websites, HR, and IT help desks to improve employee access to information and resolve open issues more efficiently. This helps educate our employees how to use GenAI in a controlled and safe environment while also building confidence with both business leaders and regulators.

The long-term success of AI/ML will need to be quantifiable and provide tangible economical and societal value. That can only be achieved if we collaborate across the public and private sector to jointly deliver responsible and trustworthy AI at scale. We will need regulators, tech companies, the banking industry and client advocacy groups to establish the right adaptable frameworks and guardrails to maximise the positive impacts of AI/ML on our industries and society and minimise its negative impacts.

The element of trust is key, because that’s what FS firms are built upon. We must ensure a consistent and transparent standard that’s supported by regulators, which requires firms to talk to each other regularly. We’re already seeing approaches to AI regulation and compliance emerging across the industry to help safeguard employees and ensure better experiences for the whole spectrum of customers.

Success here relies on robust AI model testing as well as sufficient access to data. The real power of this technology is unlocked when you combine your data with publicly available data, because that massively expands your horizons. But this requires your business to have the right skills to ensure the compliance of your AI models in this highly regulated industry. Having risk, compliance and legal divisions that understand the technology and are up to speed can really help here. We must embrace these technologies, but we also need to keep customers and colleagues safe. That’s super-important in FS.

Sean’s right. We need to embrace innovation but not at the expense of customer safety. At Deutsche Bank, we’re working hard to develop appropriate guardrails and train our subject-matter experts in the responsible use of AI tools.

One example of this is the so-called GenAI omission challenge. A common use of GenAI is to summarise documents and reduce, say, 50 pages to three bullet points. But maybe the summary isn’t quite right. You ask for three more bullets, then three more and so on. But how do you determine which are the right three points to choose? That will require a ‘human in the loop’ with active subject matter expert involvement and oversight. Getting this right is essential and reaffirms the need for every company to establish appropriate ‘guardrails’ and upskill employees to ensure these types of GenAI tools become an integral part of the FS industry.

We’re going to see a kind of twofold strategy. First, FS firms will become more familiar with this technology and start to embed GenAI within the business. Testing and refining this tech internally will give them space to get things right before rolling it out to customers.

The second phase will be this customer-facing piece in which large language models become more sophisticated and enable bank staff to improve their interactions with customers so that they become more natural and intuitive.

I agree with Sean. it’s clear to me that AI/ML is improving, and will continue to improve, the productivity of everyone involved. These kinds of gains will help firms to grow, increase market share and achieve more with the same resources already available to them. You could also use these productivity gains to increase profitability and, in effect, do the same with less.

It’s for each company to decide how to use the gains that AI/ML can offer, but what is certain is that this technology, especially GenAI, will have a fundamental impact on every aspect of banking. It will redefine how we operate and interact with our employees, partners and clients – and bring some much-needed magic to the industry as a whole.