The power of AI is reshaping businesses’ operations and strategic decision-making paving the way for more investments in advanced technologies that can drive truly transformative outcomes. Adopting AI and associated solutions has become a critical necessity,
enhancing service offerings and providing a competitive edge. But despite these advantages, many businesses are still apprehensive about adopting AI.
One reason for this is the stringent and continuously evolving data privacy AI regulations that businesses must comply with. Another is the perceived lack of accuracy. Failure to deal with these elements can clearly lead to substantial reputational and financial
consequences. These concerns often lead businesses, especially those in the financial services sector, to delay the adoption of advanced technologies, which in turn hampers their progress.
Boosting AI innovation through new initiatives
Although the sector is still apprehensive about adopting AI, there’s a way to alleviate any concerns through safe experimentation and deployment. For example, providing a platform where organisations are able to test the technology in a regulated environment
will allow them to better understand the use of AI. With this approach in mind, the Financial Conduct Authority (FCA), has recently launched a Supercharged Sandbox to help firms experiment safely with AI using tools from the US company, Nvidia. This is a major
step towards increasing AI innovations while also further contributing to the UK economy.
With the supercharged sandbox, companies have the opportunity to safely check advanced AI technologies under the regulators watch. These tools have – for example – the power to identify fraud and track any suspicious activity, allowing those that do not
have the capabilities on their own to explore and understand the processes. As the UK hopes to lead in AI transformation, this programme is an opportunity to encourage companies to explore new ways of using this technology.
But will access to such a technology drive the sector’s anxieties away? Not entirely, as just having a platform is not enough. Long term success will require expert change management, continuous post launch reviews and transparency in tracking and analysis.
Challenges while exploring AI tools
While this initiative provides an opportunity to experiment with AI, no institution is going to deploy the technology in the real world without absolute certainty about its accuracy and robustness. Creating in the sandbox, an AI app that is 95% effective
at detecting fraud might not be good enough when you must accept 5% of the cases will be false positives. The question is not whether 5% is okay, but is it better than the current human and systems solution? Banks will also look to replicate to a large degree
the policy frameworks, sign off and escalation levels they have in place today. This will include therefore some level of human oversight built in as a check and balance and at a point where the bank or regulator deems QA is needed.
Moreover, it’s crucial to note that these AI tools will be built into the banks’ risk appetite, policies and regulatory frameworks. For example, where solutions touch on processes like claims or lending, then banks have to be able to explain and show that
the right decisions have been made. This needs explainability, transparency, auditability and the ability to escalate for review.
Building transparent and predictable AI
It’s key to recognise that once AI is deployed, it must be regularly monitored and ensure a transparent decision making process to safeguard organisations from any errors or systematic bias. Think of how today any bank monitors the decisions their credit
risk staff make to ensure they align with the bank’s lending policies. This is no different in an AI driven world.
Transparency and predictability in AI processes will encourage customers and employees to further engage and use these operations. For instance, one way to achieve predictable AI is for workflows to be built with rules and structure, ensuring they are aligned
with the company goals, eventually setting them up for success. These agents and AI are monitored and easily interrogated, all to ensure predictability and transparency, which after all is what a certain element of bank trust is based upon. With AI rooted
into workflows as bank standard processes, it will allow companies to be more efficient across departments while ensuring minimal errors especially in such regulated sectors.
Ultimately, as AI becomes more tested and adapted to improve processes, its impact on cost and productivity will be huge in operations and customer service and a scale of change like the impact offshoring had in the 90s and Noughties. Across many areas this
will be a 50%+ change. This will require disciplined execution and change management given the impact on roles and ultimately the end customer.
