Generative AI and payments: The brave new future?

While AI as a whole is revolutionizing the financial sector by automating processes, enhancing accuracy, and cutting costs, Generative AI’s unique contribution lies in its capacity for innovation and creativity. Generative artificial intelligence (GenAI) stands out as a remarkable facet of AI, primarily known for its ability to generate new content (ranging from images and music to comprehensive texts and even code) based on learned patterns trained through machine learning on vast datasets. 

Generative AI is a significant milestone in the evolution of artificial intelligence and as such, and it will impact not only products and services, but also different functions in organizations and the levels of productivity. Nowadays it represents only a fraction of all artificial intelligence use cases. However, according to McKinsey’s research, GenAI will have a significant impact on all industry sectors and its impact on productivity could potentially contribute an additional $2.6 trillion to $4.4 trillion annually, amplifying the overall impact of artificial intelligence by up to 40%.1 

What does generative AI mean for payments?

While financial institutions are already harnessing artificial intelligence for tasks such as cash flow forecasting, credit scoring, fraud prevention, and compliance, the introduction of generative AI promises to broaden the spectrum of applications even further. Productivity increases due to generative AI use cases have the potential to create a revenue increase of 2.8 to 4.7% for the banking industry. Furthermore, integrating generative AI with other AI tools can help increase customer satisfaction and reduce risk. 

In fraud prevention, for example, the ability of AI to process vast data volumes in real-time is invaluable. It aids in detecting and preventing fraudulent transactions, allowing payment providers and financial institutions to distinguish between “good” and “bad” payments. While generative AI can contribute to understanding patterns, it’s the combination of various AI models and old fashion rules that plays a crucial role. And, effective fraud detection not only allows you to stop treating all customers with suspicion, reducing friction and frustration, but also results in cost savings. 

Moreover, by analyzing consumer behavior and spending habits, AI algorithms can offer personalized product and service recommendations. This paves the way for banking-as-a-service or loyalty-based products, enhancing the consumer experience and generating additional revenue for banks. 

Potential use cases

Given the banking industry’s distinctive blend of expertise, technological advancements, diverse customer interactions (both digital and in-person), and the fact that it’s immersed in a rigorous regulatory framework, it stands as an ideal platform for applying a wide range of generative AI use cases. Let’s delve into some of these potential use cases for generative AI in financial services. 

Customer service 

Chat bots enhanced by Generative AI are evolving rapidly. These advanced chatbots can generate contextually relevant responses and offer tailored recommendations, going beyond the capabilities of traditional chatbots. By harnessing the power of generative AI, they can address a wider range of customer queries, leading to reduced wait times and improved satisfaction levels. 

Risk management

Financial institutions can improve risk management with AI by identifying potential issues and providing early warning signals. By analyzing large amounts of data, AI can help to identify patterns and trends that may not be apparent to human analysts, helping banks to make more informed decisions and reduce their risk exposure. Furthermore, generative AI can be employed to visualize and interpret complex patterns, enhancing the understanding of other AI models and providing deeper insights into data-driven decisions. 

Fraud detection

With the capability of analyzing vast amounts of data, AI models can discern patterns of “good” behaviour. Properly implemented machine learning algorithms, trained to recognize standard transactional behavior, can quickly approve and accept the vast majority of transactions, preventing innocent customers from being impacted by friction generating additional checks and steps. This allows more time and effort to be spent on dealing with “bad” transactions, putting in additional checks, authorisations and approvals, in order to deter and prevent the bad actors. Generative AI, in combination with other models, can simulate various transactional scenarios, helping banks better understand ”normal” patterns and refine their fraud detection mechanisms. 

Challenges and Ethical Considerations

While the benefits are numerous, there are inherent challenges and ethical considerations that financial institutions must navigate to fully harness the power of this technology. Here are some: 

• Infrastructure: As AI models become more complex, they demand significant computing power. Financial institutions must ensure they have the necessary infrastructure in place to support these advanced models. The challenge lies not only in having enough processing capability but also in ensuring scalability and efficiency as data volumes and model intricacies grow. 
• Data quality: AI relies on large amounts of data and generative AI is deeply tied to the quality of the data it’s trained on. Therefore, financial institutions must ensure the accuracy and reliability of their datasets. 
• Data privacy: In the era of generative AI integration, customer privacy is critical. Financial institutions must prioritize not just compliance but also the trustworthiness of their systems to ensure that AI deployment doesn’t endanger data confidentiality. 
• Regulatory compliance: Adherence to regulatory standards is crucial. Institutions must navigate the evolving landscape of AI regulations, ensuring that their implementations align with both current and emerging standards. 
• Reliability: As AI becomes integral to financial operations, its reliability becomes non-negotiable. Financial Institutions must ensure that AI systems consistently perform as expected, minimizing any errors and maximizing uptime. 
• Transparency: All decision-making processes, especially those supported by AI, must be clear and comprehensible to customers and stakeholders, ensuring AI’s operations and outputs are understandable to all. The sharing of the reasons for the decision-making that has taken place is becoming expected and could soon be regulated. 
• Fairness: Generative AI systems must be designed freely to avoid bias and discrimination based– whether related to factors such as race, gender, or socioeconomic status. It’s not just about ethical AI, but about creating systems that are equitable for all users.  Being able to show that models have been checked for bias and that data points that could cause a group to be disadvantaged are not being used is good practice, and should not be the exception when deploying AI. 
• Accountability: Financial institutions are responsible for the decisions made by their generative AI systems. They must be able to explain, justify and, if necessary, correct the outcomes generated. 

How are financial institutions already harnessing Generative AI?

Morgan Stanley is piloting a chatbot, powered by OpenAI’s GPT-4 language model, to enhance the service offerings of its financial advisors. This advanced chatbot can address queries about financial offerings, produce reports, and assist in research. While the initial testing phase involves a select group of advisors, Morgan Stanley intends to expand its usage soon. Morgan Stanley (openai.com)

Fujitsu and Mizuho Financial Group have initiated joint verification trials using Fujitsu’s generative AI technology to enhance the development and maintenance of Mizuho’s systems. The collaboration aims to harness the transformative potential of generative AI in optimizing work processes, improving system quality, and bolstering resilience during the development and maintenance phase. Fujitsu and Mizuho start trials for generative AI to streamline development and system maintenance operations: Fujitsu Global