Artificial intelligence (AI) can be a powerful but complex asset for financial services businesses, but its successful deployment involves not just capturing operational efficiencies but also embracing ethical responsibility and environmental consciousness. That was the message of a recent Talking Financial Services webinar, ‘AI: Risk vs Reward’, featuring ACCA member and virtual finance leader Conrad Nevers and Lloyds Banking Group’s Dr Nataliya Tkachenko.

RPA vs AI

The webinar made the point that AI is often mistakenly equated with robotic process automation (RPA). There is, however, a crucial difference between the two technologies. While RPA can be given a set of preset rules in order to automate repetitive tasks, AI is a much more powerful tool that can be trained to interpret data, identify patterns and make data-driven recommendations. This gives AI a distinct edge in high-stakes decision-making, such as fraud prevention and credit scoring, where its adaptability and ability to learn from data are crucial advantages.

AI’s application within the finance sector is evolving from experimental to essential. Its ability to handle such complex tasks as credit risk assessment and fraud detection is invaluable, allowing financial institutions to operate with greater precision, efficiency and scalability. By allowing financial institutions to leverage data-driven insights in real time, it improves the customer experience as well as operational efficiency. Its continuous learning capacity is especially valuable in managing dynamic and high-risk environments.

Sustainability

The sustainability issues associated with AI were a major talking point at the webinar. AI can consume vast amounts of energy, and the discussion made clear that firms need to understand the various layers that AI spans – from infrastructure to user interaction – if they are to implement effective carbon accounting, especially as AI models become larger and more complex.

When procuring AI tools, financial institutions should ensure any purchase decision is aligned with decarbonisation goals and support organisational and societal values. This will be the most effective way of mitigating any reputational and operational risks.

One example given was that not all financial tasks require large, energy-intensive AI models. The targeted deployment of right-size AI for specific tasks (eg smaller models for text classification) will reduce unnecessary energy use without any impact on accuracy. This approach not only supports environmental sustainability but also optimises process efficiency.

Ethics

While AI has ethical challenges – the perpetuation of bias in training data, the privacy of confidential information, and the transparency of its decision-making – the technology can also be a driver of ethicality. Nevers and Tkachenko pointed out that one AI use-case is the identification of greenwashing – the practice of companies falsely claiming environmentally responsible practices – which is increasingly seen as a form of financial fraud. Used as a tool to verify environmental claims, AI promotes transparency and accountability within finance.

Another webinar message for financial services businesses was to avoid overreliance on specific AI providers, as ‘sticky’ relationships may lead to the loss of accumulated learning following any switch in supplier. Institutions must weigh up the advantages of established relationships with large providers against the potential flexibility and innovation offered by smaller suppliers, and factor in cost, learning transferability and scalability considerations.