The cautious AI revolution: How regulated industries can safely harness generative AI
Due to strict rules governing data, highly regulated sectors like financial services, healthcare, and public sector agencies cannot simply rush headlong into deploying generative AI (genAI). However, the potential for increased efficiency and productivity is simply too great to ignore. The question isn’t whether these sectors should leverage genAI, but how they can do so while maintaining regulatory compliance and managing risks.
In some ways, this is a catch-22 situation: organisations like banks, insurers, healthcare providers and governments sit on troves of the types of data that lend themselves well to genAI projects, and often they have the technology resources to make them successful. On the other hand, much of this data is personally identifiable information, rightly subject to stringent privacy laws due to the potentially severe consequences of misuse or mishandling.
As a result, regulated sectors have not only more, but more closely watched, regulations to abide by. But with a thorough understanding of the risk landscape, strong data foundations, and well-conceived projects, regulated industries can navigate regulatory complexity and bring innovations to market that will improve their operations and stakeholder’s experience.
Let’s explore how this can be achieved.
Understand which regulations may apply to project data
Beyond technological advancement, adopting genAI in regulated industries brings with it a host of challenges that keeps the sector on high alert. While it has garnered significant attention, the inceptive nature of genAI governance and the risks associated with its deployment such as data security, ethical implications, transparency and accountability can baffle business leaders.
Trained on large data sets that include personally identifiable information as well as intellectual property, large language models may have access to data protected by various privacy regulations. Mishandling this data could lead to steep fines and reputational damage. In fact, the average cost of data breach in Australia is reported, leaving regulated industries with no margin for error.
Beyond data privacy lies another challenge: the “black box” nature of genAI algorithms. Even though AI, and particularly generative AI, can produce remarkable results, it is often difficult to understand how the algorithms lead to the outputs they deliver.
That lack of transparency can raise red flags on the reliability and fairness of genAI powered decisions, especially in domains like healthcare, finance and government, where the stakes are high. Closing this transparency gap remains as an essential step to harnessing the transformative power of genAI.
Make sure your data estate is in order
The first step to any successful genAI project is getting the data in order. Governed by strict compliance requirements around data privacy and security, a well-architected and optimised data estate becomes a vital prerequisite for regulated industries. Data should be pristine — cleansed and categorised. This can be easier said than done, with legacy data silos, outdated access controls and murky data lineage presenting significant roadblocks.
Venturing into data modernisation initiatives can become a critical aspect to successfully leverage genAI solutions. Migrating from legacy on-premises systems to modern data infrastructure can ensure not just data quality but its accessibility, scalability and security — bringing a welcome order to the chaos.
But modernisation alone isn’t enough. Organizations will need a comprehensive data governance framework that tracks data lineage, enforces airtight access permissions and enables responsible data sharing, all while staying inside the lines of the regulatory boundaries. While this process can be time consuming and costly, it proves to be a crucial investment for regulated industries looking to harness genAI safely and effectively.
Be strategic with project selection
Despite challenges, forward-thinking organisations in regulated industries are finding ways to leverage genAI safely and compliantly. The key lies in identifying use cases with the best risk-reward ratio. This often means starting off with smaller projects that can serve as a testing ground.
For instance, a healthcare provider may have a vision of using AI to automate large swathes of the patient experience. But starting with an initial project, like using GenAI to assist doctors in generating case summaries , creating specialist notes, and improving documentation accuracy, may be the best approach, as it can be relatively fast to market and present minimal risk.
Organizations should take a long view when strategically planning out how they can use AI to innovate, and build a roadmap that starts with lower-risk projects. This roadmap should be evaluated frequently and refined based on the learnings from each project.
Charting the path forward
As AI — and particularly GenAI — continues to evolve, regulated industries have a unique opportunity to leverage technology for competitive advantage. By carefully navigating the regulatory landscape, getting their data estate in order, and focusing on the right use cases, these sectors can drive innovation while maintaining compliance.
The journey may not always be easy, but the potential for increased efficiency and productivity, as well as superior user experiences, makes it worthwhile.
