Whilst advances in technology have helped trigger the explosion of data, technology also provides the solution for harnessing it. This extends to the work of investigators and compliance professionals. Artificial intelligence (AI) and machine learning are intuitive technologies with many investigative and compliance applications, which are increasingly embraced and encouraged by regulators as well. Despite common misconception, however, human input remains essential in the design and day-to-day implementation of these technologies.
FRA Director Weng Yee Ng and Manager James Norden set out practical guidelines, benefits and challenges of embracing AI and machine learning in increasingly data-heavy investigations and compliance programmes.
AI and machine learning technologies are already being deployed in corporate investigations. Examples include technology-assisted review (TAR) for narrowing down document reviews and AI solutions for configuring or eliminating sensitive materials based on jurisdiction-specific requirements. However, any AI methods used in investigations must be sufficiently robust and defendable in legal proceeding and carefully considered in relation to alternative information sources available to regulators.
Continuous compliance monitoring also benefits from AI and machine learning, especially where compliance teams are small or required to comply with multi-jurisdictional regulations. Deployed effectively, AI can enable greater geographical coverage, consistency and efficiency of approach, whilst being defendable and facilitating real-time detection of issues requiring remediation.
Among the many benefits of deploying AI in investigations and compliance programmes is the ability to allow different information types and siloed data sets to ‘speak’ to one another, in the interest of improving sorting and analysis. AI also enables long-term cost savings by automating certain investigation and compliance processes and focusing human resources on making key decisions based on the output. Furthermore, the systematic removal of sensitive information (e.g. personal data) facilitates compliance with increasing data privacy laws. Overall, accuracy and efficiency improve with the use of learning algorithms that are designed to repeatedly fine-tune themselves to make output more relevant and reduce the potential for human error.
AI may not be appropriate where it is not fully understood, where insufficient data is available, or where a company lacks the time to deploy it effectively. Experienced investigators and compliance professionals must be consulted to ensure data privacy laws are not breached, and the process is appropriate for the complexities of the specific task at hand.
Accreditation: An extract from the 2020 edition of Asia-Pacific Investigations Review. The whole publication is available at https://globalinvestigationsreview.com/edition/1001390/asia-pacific-investigations-review-2020