Regulatory Compliance: How to Gain High ROI on AI with Optimal Human in the Loop across all Lines of Defence

Introducing Our Author

In this blog, Shwetha Shantharam, AVP and Product Head at 4CRisk.ai, joins us to look at some interesting studies, surveys and deep dives that shine a light on what organizations can do now to ensure their AI-powered Regulatory Compliance deployments are successful. With more than 20 years’ experience, Shwetha has focused the last 5 years on powering up products for regulatory, compliance and risk teams with AI.  

Shwetha shares how empowering your front-line and second-line professionals with smart and effective Human oversight and  Engagement, your AI projects will be more successful.  People need to be able to steer the use of AI so that it enriches their roles, rather than threaten their jobs.  
‍

1. MIT’s recent study ‘The Gen AI Divide’ - What is the ’Learning Gap’ impeding successful AI powered tools and software for Regulatory Compliance and RegTech and what can we do about it?

Over the past year, we’ve seen AI projects stumble and fail: MIT claims that over 95% of Generative AI deployments fail to deliver measurable financial value – despite $30 to $40 billion in enterprise spending, the vast majority of companies remain stuck, unable to extract real value from their AI initiatives beyond enhancing personal productivity. The report claims “60% of organizations evaluated enterprise-grade systems, but only 20% reached pilot stage and just 5% reached production – typically in tech or media sectors. ​Key barriers include brittle workflows, lack of contextual learning, and misalignment with operations”.
‍

The main finding in the MIT study is that the quality of the AI models was not the impediment, but the “learning gap” for both tools and organizations, underscored by flawed enterprise integration.  The study claims, “Users resist tools that do not adapt or integrate well into workflows. ​Model quality concerns arise when enterprise tools fail to meet user expectations. ​A significant number of workers prefer human input for complex tasks due to AI's lack of memory and adaptability.”
‍

There’s been pushback to the MIT study;: The Marketing AI Institute refutes these claims, founder and CEO Paul Roetzer claims the methodology is flawed with too narrow a definition of success with ‘ROI impact measured six month post pilot.” and little accounting for crucial business impacts like efficiency gains, cost reductions, customer churn reduction, lead conversion improvements, or sales pipeline velocity.

What does the MIT’s recent study ‘The Gen AI Divide’ MIT tell us about what AI Regulatory Change Management tool Success looks like?

• Empower Line Managers and SMEs to drive adoption – make sure they understand, support and champion the real value the deployment with bring with specific use cases that your team agree needs a higher level of automation.  This means gaining consensus on process re-streamlining, involvement with the business case, including the expected ROI and validating results with a pilot.
• ‍Know how you’ll measure success as the deployment is rolling out. Look for  efficiency gains and productivity improvements and lower outsourcing costs, but also unexpected benefits – like the quality of result and professional self-esteem lifts from making a real impact on the company’s success. Be there for the long game.
• ‍AI embedded in a streamlined business process is Cultural Change and needs to be treated like a major organization change with ongoing AI education,  training and experimentation.
• ‍Use Human Oversight to tweak and improve.  A product that incorporates process steps where professionals can verify, collaborate on results, and revise is going to be more trusted by its users. Human oversight and feedback is vital and can help ensure AI-generated results are reliable, accurate and build trust.  

Best Practices Practical Tip: Get the 4CRisk eBook on determining ROI for AI Projects

2. 1LOD' 2025 1st Line Risk & Control Benchmarking Survey and the 1LOD Deep Dive Report on  Rethinking Roles, Responsibility and Risk

This recent published survey gives real insights into what is tripping up the front line.  Here’s a sample:

• 72% say the biggest challenge in implementing regulatory change within the 1st line is technology limitations for adapting controls and processes

• 70% say that data availability, quality and labelling for AI models is a highly significant barrier to delivering an effective controls platform

• 62% say that integration challenges with legacy systems and traditional controls are a significant barrier to having an effective controls platform

 
Let’s look a few of 1LOD’s recommendations:

• Modernise regulatory change management – move from procedural to technology-led compliance.

• Improve resources for horizon scanning – create a dedicated, tech-enabled capability to spot emerging threats

• Automate controls and unify tooling in control assurance and issue management – reduce the manual load in business-as- usual (BAU) processes,  

• Integrate governance, risk and compliance (GRC) and surveillance systems – enable seamless linkage between risks, controls, incidents, and regulatory changes.

• Embed AI and predictive analytics – shift from pilots to operational tools in surveillance, risk scoring, and elsewhere.

In 1LOD’s Deep Dive Report on  Rethinking Roles, Responsibility and Risk, we see some concrete recommendation on how to be more proactive, change the culture while deploying technology, and owning the risk

• From patchwork to proactive: build real-time control intelligence Continuous control measurement, insights driven by Artificial Intelligence (AI), and real-time monitoring are replacing reactive, manual detective controls as the new standard

• Culture first, technology second True control transformation is driven by mindset shifts and collaboration across the 3 lines. Technology enables, but culture sustains.
  

• Own the risk, not just the process Risk functions must move beyond process ownership to strategic enablement, embedding the thinking or strategy about risk directly into business decision-making from the outset.

Best Practices Practical Tip: Get the 4CRisk eBook AI Agents and Copilots for the Front Line

‍

Summary

Leveraging AI-powered Regulatory Change Management will help you keep your RegTech stack AI-future proof – but only if the deployment is embraced by your teams and  your organization can reap its value. AI helps your team keep pace with the velocity of change across all applicable rules, regulations and laws (RRLs) while mitigating compliance risks by aligning policies, procedures and controls with required changes. Insights from AI analysis reveal trends, inform strategy, improve time to market, competitive advantage and differentiation.  To be truly successful, strong ROI, team champions and human oversight that encourage engagement and continuous improvement, rather than resistance, is table stakes for success.

Check out these related blogs and resources  

• https://www.4crisk.ai/post/five-ways-regulatory-change-management-istransforming-with-ai

• https://www.4crisk.ai/post/regulatory-intelligence-how-the-regtech-sector-is-being-transformed-by-ai-in-regulatory-risk-and-compliance-programs

• https://www.4crisk.ai/post/what-leaders-should-consider-when-buying-ai-powered-regulatory-change-management-solutions
  

‍