Hot Topic: Artificial Intelligence

Don’t automate the humans: AI for financial services.

Artificial intelligence (AI), especially in the banking world, is hitting all sorts of headlines lately. But did you know that there is an actual discipline that studies the interaction between humans and technology, and how technology affects culture? It is called cyborg anthropology: most people nowadays being defined as cyborgs due to the very fact we interact with technology.

The SWIFT Institute spoke to Amber Case, a cyborg anthropologist and author of Calm Technology: Principles and Patterns for Non-Intrusive Design about the impact of AI on the financial services industry. We also spoke to Edouard d’Archimbaud, Head of the Artificial Intelligence Lab, from the Corporate and Institutional Banking arm of BNP Paribas, about what financial organisations can expect out of pursuing a strategy of AI implementation.

Machine learning

Amber Case is “third-generation AI”, coming from a family of people working on artificial intelligence and machine learning. Her grandfather was an AI researcher at the University of Utah and worked on Project Mercury, the first US human spaceflight programme, and ARPANET, the predecessor to the Internet. Her father was an AI technologist for telecom companies and read to her Naturally Intelligent Systems as a bedtime story. As a result, Case much prefers the expression ‘machine learning’ to the term ‘artificial intelligence’ as she feels the latter tends to lead people astray. “Machine learning uses human information and context, instead of just raw programming,” explained Case. “I think many people think that if we advance enough in our state of technology, we will be able to automate everything and won’t have to work any more. But we have already automated a lot of things and it is those things that are giving us the most trouble right now. What we should aim for is a more cybernetic system, machine learning that works on a feedback loop between humans and technology. That is how we can achieve really smart systems that adapt and grow, and not end up being brittle.”

Case cited examples of automated systems that often fail us, such as automated parking garages when the machine cannot read your ticket, or being kept on hold for 40 minutes on an automated telephone line. “I think that as we get more automated it will become more important to have real people on the other end; instant customer service and talking with a person who understands you,” said Case. “This is the opposite of people’s assumptions – we are not going to automate humans. The best technology should automate the boring, repetitive tasks in order to allow us to be more human.”

Perceptive patterns

Because we experience time at a human pace, it can be difficult for us to see patterns over time. But technology is great at showing and helping us understand these patterns. Patterns that would be helpful to people in terms of managing finances could be, for example, tools that help understand a person’s spending patterns as an end-user with anonymous comparisons against other end-users, or a system that provides alerts when users are about to hit particular spending limits. Smartphone apps, such as Acorns, can invest the spare change from each of your retail transactions. Ideally these types of systems should work invisibly in the background for the end-user.

Intelligent innovation?

Case’s view is that it is difficult for large financial institutions to do innovative things. “Even if a bank has an Innovation group, usually the organisation is so large it can be very difficult to allow people the freedom to make something really useful,” explained Case. “I think, therefore, there is a lot to be said of acquiring the best startups in the market after they have gained traction and been listed, say, for a minimum of three to five years. Financial institutions should bear in mind that these applications can supply a really good front-end for unwieldy banking applications and transactions.”

Case also made the point that it is not always about using the newest technology, but rather about making older technology sustainable. Banks have the choice of either simplifying their systems and stitching them together, or acquiring a new platform and leaving it alone, letting it grow and allowing people to eventually adapt, or by stitching some of the new analytics from an acquisition into their base platform.

Not surprisingly, Edouard d’Archimbaud from BNP Paribas CIB disagreed with Case regarding innovation within large financial institutions. “We looked at the origins of the most successful digital companies of today such as Facebook, Google, and Apple, and basically they all started in a garage. We set up our Artificial Intelligence Lab in the spirit of going back to the garage days, meaning a small team of diverse talents with less meetings and no red tape. We develop smart applications based on state of the art machinery and artificial intelligence.”

D’Archimbaud emphasised the ethos behind BNP Paribas’ activity in automation. “AI could be the biggest technological shift in our history, bigger than either the computer, Internet or smartphone revolutions. Not only do we have to prepare ourselves for it, our clients expect it. Furthermore, banks are no longer responsible for simply managing financial transactions on behalf of their clients, they are also responsible for processing and safeguarding their information.”

At your service…

We asked if there were some banking systems that were more suitable for AI than others. BNP Paribas identified four main areas including customer experience, risk management, operational efficiency, and inventing new business models. In general, robotics and AI are expected to be able to automate some of the boring, repetitive, labour-intensive tasks performed by bank employees, so that they can focus on more tasks that require expertise and creativity. Overall, BNP Paribas wants to enhance the client digital journey and automate some of the answers to the simplest of questions so that staff can focus and give more value on the more difficult queries.

Smart, personal, conversational assistants, multi-language reporting, KYC-compliant onboarding – BNP Paribas predicts that all of these will become the new standards with the expectation of greatly improving the customer experience. BNP Paribas looks forward to implementing these new technologies and use today’s workforce to instead build more new products that will generate higher value for their clients. The focus is to continue to move forward BNP Paribas’ business goals by increasing the use of innovation.

Case reminded banks of the importance of keeping integrations open across platforms, stating: “I think the importance of these systems is that if you allow data to be shared and imported and exported really easily, people will make really great interfaces for them. Not only could a bank source these startups and services as potential acquisitions, but it would also increase the likelihood of startups using the bank’s software because they would be able to do more with the bank’s data.”

Ethics of AI

Regarding the subject of ethical challenges for banks when adopting AI, issues such as data protection remains one of BNP Paribas’ most utmost concerns. One of the reasons the financial industry has been perceived to be moving slower than some startups in the FinTech world is because ultimately traditional banks need to ensure protection of the status of trust gathered from clients over many years.

Likewise, Case also highlighted the importance of trust regarding the protection of data. “Organisations have a fiscal, ethical and human response to protect people’s data,” said Case. “Also, when firms share data with third-parties, they need to say it in plain English and ensure that data is protected by those third parties.”

The human touch

Will AI ultimately help society and individuals move forward? D’Archimbaud commented: “In order to get the best out of AI you need to have it combined with a human. AI is like Google Search – it makes searching for information much more effective, allowing us to focus on added-value tasks where creativity is paramount.”

Case agreed saying, “As long as we include humans in it, we’ll be just fine. But if we try to automate humans out of the system, we will end up turning into robots ourselves. Our tasks will be supervising machines instead of creating. Everybody wants to call tech support and have a real person talk to them, because at the end of the day people want human connection. Technology will always have edge cases, and we need humans to get out of them.”

D’Archimbaud concluded, “I think AI will make the world more efficient, creating a better environment for our clients, employees, shareholders and regulators. I am very positive. I think it will basically create a better world, one in which improved integration of technology could actually make the world more human.”