Artificial intelligence (AI) is transforming the landscape of Regulatory Technology (RegTech), offering unprecedented opportunities to improve compliance and manage risk more effectively.

However, with such rapid advancements come significant challenges, especially surrounding the secure and responsible use of AI. In this article, we’ll explore how AI can be effectively integrated into RegTech. Along the way, we’ll address the crucial questions of when and why to use AI, and whether we are truly better off with it.

Why Are We Using AI?

Before delving into the NCSC guidelines, it is essential to ask: why are we using AI? For Jason Maude at Starling Bank “We’re using AI where it makes sense to use AI”¹. But what does this actually mean? It could be that it reflects a growing realisation that AI should not be adopted any which way but should be treated like a tool and used where appropriate to solve specific problems and improve efficiency.

Traditionally, financial institutions have been data rich but not necessarily data centric. This shift towards data-driven decision making is where AI becomes transformative. With mountains of regulatory data, and the need to meet complex standards, AI helps make sense of this information more effectively and efficiently than the human brain could.

However, it’s worth noting that not all AI implementations are successful. AI models require the best data engineers to utilise the most accurate pseudo data for a useful model to be created. Unfortunately, this isn’t always the case with 8 out of 10 AI platforms failing, due to poor planning, lack of data quality or insufficient integration with business processes. It is for these reasons that AI needs to be carefully considered before being adopted in RegTech.

Do We Need to Use AI?

While AI offers benefits, we must also ask: Do we need to use AI for every regulatory task? The answer is not always straightforward. There are scenarios where traditional methods and simpler technologies will suffice. For any compliance process that involved nuanced human judgement, empathy, or complex ethical considerations may not be ideal for AI automation.

There is also the behavioural science perspective. There are distinct differences between human to human and human to AI interactions. Humans respond more positively to empathetic, human led interventions compared to interactions with AI. The emotional intelligence and adaptability of human compliance officers cannot be replicated by AI. AI excels at automating repetitive, data-heavy tasks, but it cannot replace the human element. A study showed that humans interacting with AI are much more likely to lie, with 62% lying to an AI bot, compared to 12% lying to a human.

Are We Better Off Using AI?

The final question we must grapple with is, are we better off using AI? To answer this, it’s important to weigh AI’s advantages against its potential drawbacks. AI has proven to be transformative in automating regulatory processes, reducing human error, and handling tasks at scale. It can also learn from data, making it highly adaptable to an organisations unique workflows and regulatory requirements.

However, AI also comes with significant risks. One of the most pressing concerns is data quality. AI models are only as good as the data they are trained on, the input directly translates to the output. If the data is flawed, biased, or incomplete, the AI’s outputs will be problematic. Ensuring high-quality, unbiased data is crucial, but it is also resource-intensive and requires continuous monitoring and oversight.

There are also ethical considerations. The risk of AI models learning harmful behaviours from external data, such as the Twitter bot, serves as a cautionary tale. RegTech AI systems must be designed and monitored to prevent such failures. Overreliance on AI poses risks, as human judgment and oversight are essential in interpreting regulations.

Despite these concerns, there are many areas where AI has excelled is making it easier to detect fraudulent activities and manage compliance in real time. The question of whether we are truly better off hinges on how well AI is integrated into the broader risk management strategy and whether its use aligns with long-term goals.

A Delicate Balance

The responsible use of AI in RegTech and beyond requires a delicate balance between innovation, efficiency and caution. Firms must be thoughtful in their application of AI, using it where it adds genuine value. However, they should remain vigilant about the technology’s limitations, particularly in areas requiring high data quality, transparency and human oversight.

Ultimately, the decision to use AI should be driven by a strategic assessment of whether it genuinely improves outcomes or simply adds complexity. As AI continues to evolve, continuous evaluation and responsible governance are paramount.

Discover how Axiol can transform your compliance processes by booking a demo today. Visit our website or contact us at info@axiol.com for more information and to explore how we can help you achieve seamless, efficient regulatory compliance.¹ Jason Maude, Chief Technology Advocate, Starling Bank, speaking at FinTech London 2024