Banks and the wider financial services sector are poised to benefit from savings arising from generative artificial intelligence (AI) to the tune of $200bn to $340bn annually, according to a report published by McKinsey in June.
On top of that impact, the use of generative AI tools could also enhance customer satisfaction, improve decision-making and employee experience, and decrease risks through better monitoring of fraud and risk.
According to the global management consultancy, through the deployment of AI the banking industry has the potential to improve on efficiencies already delivered by AI by taking on lower-value tasks in risk management, including required reporting, monitoring regulatory developments and collecting data.
The report cites one European bank that has leveraged generative AI to develop an environmental, social, and governance virtual expert by synthesising and extracting from long documents with unstructured information. The model answers complex questions based on a prompt, identifying the source of each answer and extracting information from pictures and tables.
Digital start-up banks and traditional banks are each in a similar position to take advantage of the opportunities presented by generative AI, according to Carlo Giovine, a partner at QuantumBlack, the centre for advanced analytics and machine learning for McKinsey, which produced the report.
He says both kinds of bank have particular strengths they can leverage to capture value from this form of AI.
Time to get started
Mr Giovine says banks should not hesitate to assemble a generative AI strategy. “My advice to banks is they need to get started and therefore it is easier to start with what they have already in place, for example a cloud provider that has already been on-boarded through banks risk and info security processes,” he says.
But he advises they also need to work back from the high-value-delivering problems. “That’s where they learn the most valuable lessons,” he adds.
Banks have already begun to grasp the potential of generative AI in their front lines and in their software activities, the report notes. Early adopters are harnessing solutions such as ChatGPT, it says, as well as industry-specific solutions, primarily for software and knowledge applications.
The report also suggests that, in addition to increased revenues, generative AI could have a significant impact on the banking industry, generating value from increased productivity of between 2.8% and 4.7% of the industry’s annual revenues.
The use of generative AI tools could also enhance customer satisfaction, improve decision-making and employee experience, and decrease risks through better monitoring of fraud and risk, the report says.
Some banks have under-invested in data scientist or data engineers
The rapid evolution of generative AI is causing a headache because stakeholders are grappling with generative AI’s impact on business and society but without sufficient context to help them make sense of it, according to the report.
The speed at which generative AI technology is developing is not making this task any easier. It says ChatGPT was only released in November 2022, and four months later OpenAI released a new large language model, GPT-4, with markedly improved capabilities.
Banking, which the report describes as “a knowledge and technology-enabled industry”, has already benefited significantly from previously existing applications of AI in areas such as marketing and customer operations. Generative AI applications could deliver additional benefits, especially because text modalities are prevalent in areas such as regulations and programming language, and because the industry is customer-facing, with many business-to-consumer and small-business customers, it says.
Banks will have to invest more to fully capitalise on the advantages of generative AI, says Mr Giovine.
“Even when you have all the right elements in place such as strategy, data and technology and agile ways of working, you still need to push those technologies or use cases to adoption, which is the way you capture value. Some banks have under-invested in data scientist or data engineers, for example,” he says.
Characteristics that position the industry for the integration of generative AI applications include sustained digitisation efforts along with legacy IT systems, large customer-facing workforces, a stringent regulatory environment and white-collar industry.
Generative AI’s impact could span the organisation, assisting all employees in writing emails, creating business presentations and other tasks, says the report.
Generative AI has the potential to reduce the significant costs associated with back-office operations. Customer-facing chatbots could assess user requests and select the best service expert to address them based on characteristics such as topic, level of difficulty and type of customer.
Through generative AI assistants, service professionals could rapidly access relevant information including product guides and policies to instantaneously address customer requests.
Potential problems
The report warns banks that they need to be mindful of four factors, starting with the level of regulation for different processes. This can vary from unregulated processes such as customer service to heavily regulated processes such as credit risk scoring.
The second factor is the type of end user. End users vary widely in their expectations and familiarity with generative AI: for example, employees compared with high-net-worth clients.
Read more on AI
Third, McKinsey says banks must take into account the intended level of work automation. AI agents integrated through application programming interfaces could act nearly autonomously or as co-pilots, giving real-time suggestions to agents during customer interactions.
Finally, banks must consider data constraints, says the report. While public data such as annual reports could be made widely available, there would need to be limits on identifiable details for customers and other internal data.
Banks have been investing in technology for decades, accumulating a significant amount of technical debt along with a siloed and complex IT architecture, the report concedes.
The rate of acceleration of generative AI is rapid, according to Mr Giovine. “In 2017, the earliest scenario for when technology would reach average human performance in understanding natural language was expected to be in the late 2020s,” he says. “But now we can see that this level of performance is already available today.”
