One robot can do the work of up to five human employees, according to Capgemini. A Udemy study says 43 percent of workers cite the fear of losing their jobs to AI as the main cause of stress at work. On the other hand, Gartner is predicting that AI could create 2.3 million jobs by 2020, exceeding the 1.8 million that it could wipe out.
Regardless of the impact on job vacancies, AI in banking is a fast-developing reality. The cost of developing and deploying systems for specific uses in the workplace and adaptation to bank's legacy portfolios is influencing the pace and extent of automation take up.
Moreover, sector organizations are faced with the task of determining quality and quantity of workloads as the technology is introduced. All this should ultimately translate to the holy grail of improved customer service and workforce efficiency.
Cutting down on time-wasting
One of the current and most basic objectives for AI at banks is a reduction in time “wasted” on any task that can be automated. AI technologies are serving banks well in the realm of robotic process automation (RPA) for rules-based tasks, such as entry, validation, and manipulation of data, as well as creation, uploading, and exporting of data files.
Account reconciliation, report generation, mortgage approval, notification of delinquent loans and audit support are other examples of back-office activities to which artificial intelligence can be applied.
According to Chuck Monroe, head of artificial intelligence enterprise solutions in the innovation group of US bank Wells Fargo, AI is improving banking processes by providing the ability to sift through data quickly and extract valuable insights, with people still being a key part of the equation. He says:
It’s an exciting time for team members because AI can cut down on some of the monotonous tasks while allowing team members to focus on more complex tasks and have access to information that helps them do their jobs better.
At Wells Fargo, artificial intelligence systems are helping bankers in various tasks such as assisting customers in financial matters following a family member’s death.
Such processes, according to Monroe, are often complex due to the emotional stress of the customer combined with the intricacies involved in accessing dozens of documents and systems. He says:
By deploying an AI solution, we can free these bankers up to focus on the caller and their needs, and let the AI systems surface all the right questions to ask and content to review.
Reducing human error
AI work in areas such as fraud and anti-money laundering demonstrates that banks also want to reduce margins for human error. That is because incumbents are having to contend with increasingly sophisticated criminal activity, tighter regulation, and bigger fines for non-compliance, according to Daoud Fakhri, principal analyst at GlobalData.
The earliest AI implementations in the field of fraud detection were rules-based, where transactions are deemed fraudulent if they satisfy pre-defined parameters such as high-risk geographies and transaction values. But as Fakhri points out, this is quickly changing:
Criminals can easily circumvent traditional safeguards by using bots to set up thousands of accounts and test the boundaries of the systems to deduce the parameters and hence avoid them. Further, the rigidity of a rules-based approach means that false reject rates (FRRs) can be high.
Rules-based strategies are the most established form of fraud detection, but as Fakhri points out, they suffer from several weaknesses, including the fact that they are essentially reactive due to being based upon past attack patterns. He says:
These strategies are also labor-intensive, requiring human input to identify new attack patterns and enter these into the systems. AI-based approaches are starting to address these shortcomings.
Reducing the need for human input means that modern fraud detection systems can respond to new fraud techniques far more quickly, which according to the analyst, would help reduce losses while at the same time decreasing FRRs.
Current leadership challenges faced by leaders introducing AI as part of so-called “digital workplace” programs include to developing their own understanding of the technology potential, being able to foster a culture of learning and education about what AI can and can’t do, according to Wells Fargo’s Monroe. On how the bank is tackling the issue, he says:
We’ve created an artificial intelligence enterprise solutions team to truly organize around AI. We believe AI will soon touch every part of our business, so it’s important that we’re able to work quickly and implement solutions as seamlessly as possible.
The team works with project leaders across the enterprise to discuss potential solutions that incorporate AI and accelerate time to market by helping to align projects with core AI capabilities and reduce duplication across the enterprise.
As these projects progress, Monroe says that many jobs will be enhanced in response to emerging technology being used as an aid to human intelligence. But the executive adds education on how AI-generated information can enable customers and staff to make better decisions is paramount. He says:
As AI is implemented, it’s important that stakeholders have access to the right tools and training to see how AI can help them rather than seeing AI as a competitor or threat.
For now, it’s impossible to ignore AI’s perceived threat on people’s jobs over any notion that employees might use it to serve their employers better. Mark St John Qualter, head of artificial intelligence, commercial and private banking at Royal Bank of Scotland confirms this:
People will be inherently suspicious about the [AI] technology as they feel it will ultimately replace them.
The RBS executive echoes Monroe’s thoughts that business leaders will have to present the technology in a way that conveys its benefits as a positive transformation of the employee experience. He says:
It is critical that the leadership in the business positions [AI] appropriately to ensure that adoption does not become demotivating and a major distraction in the organization. For me this is about explaining sustainability and competitive advantage.
The issues circle around the older technology that many banks have and the layers of governance and bureaucracy that will have to be overcome to take proof of concept work to production effectively.
Legacy might make it more complex to banks to dovetail valuable customer data together with modern systems and networks, but as in every IT-led business change process, incumbents should not rely on the technology alone to implement AI successfully. The executive concludes:
AI should be a response to the strategy of the organization and a way to complement and deliver it. I think that organizations need to face up to this in a cross-functional way, with strategic engagement as it is not an ‘IT project’.
For now, organizations are presenting AI as something that enables humans to do a better job by adopting the "centaur" model — that is, humans and machines complementing each other's capabilities — which would, in theory, deliver a better result than work done by humans or machines alone.
"Workforce augmentation" is another term that is frequently used by AI enthusiasts to describe the benefits the technology can bring to organizational efficiency. How can a business describe this in a different manner without causing panic, I wonder?
But AI technology now has the power to tackle tasks that until recently were thought to be exclusive to humans - and that is not just about providing an account statement or asking for a card to be canceled over the phone. Machine learning and natural language processing are allowing advances such as algorithms that optimize themselves and bots that interact with customers in a nearly-human manner and offer investment advice based on people's financial circumstances.
It would not be unfair to say that the world is in a race to the bottom for wages and corporate spending - which suggests that if in its early stages, AI is already drastically reducing staff numbers in call centers, further advances will considerably reduce the scope of jobs (or their very existence?) as well as required skill set, further up the food chain.
Depending on how they are managed, further advances in automation could block employment prospects if organizations don't know, or cannot, reskill people to keep up with these developments. It is a tall order, but finding responsible ways to transition workforces to the digital economy should be a priority of every company adopting AI - and banks as key investors in technology should set the example.
AI can indeed be used in the same way mechanization complemented human capability in the first industrial revolution, but times are now different. As WEF chairman Klaus Schwab wrote in his latest book The Fourth Industrial Revolution - where he describes the economic benefit of current automation and data exchange trends in manufacturing technologies - it might also lead to a dehumanizing dystopia.