Over a couple of blog posts, I am exploring topics raised in a recent Harvard Business Review (HBR) article, entitled “How to Design an AI Marketing Strategy: What the technology can do today—and what’s next”, by Thomas H. Davenport, Abhijit Guha, and Dhruv Grewal where the authors focus on the use of Artificial Intelligence (AI) in marketing. I believe their work could be adapted for the compliance professional. Yesterday, I used the article as a jumping off point about how Chief Compliance Officers (CCOs) and compliance professionals can use AI for internal compliance communications and communications with key stakeholders outside your organization that you need to work with on compliance, such as third-party agents and vendors in the Supply Chain. Today I want to consider the framework that a compliance professional can implement to use these tools effectively for both internal and external marketing of a corporate compliance program.
The authors posit that AI can be categorized according to two dimensions: intelligence level and stand-alone or integrated platforms. Further, the intelligence level can be broken down into two subgroups: task automation and machine learning. Task automation performs “repetitive, structured tasks that require relatively low levels of intelligence.” They bring a level of ease as they are “designed to follow a set of rules or execute a predetermined sequence of operations based on a given input” However, such tools cannot handle complex problems such as nuanced employee requests for information. Chatbots fall into this category.  Such tools can provide basic assistance to employees during basic interactions, moving employees down a defined decision tree, but cannot ascertain intent, offer customized responses, or learn from interactions over time.
With machine learning, “algorithms are trained using large quantities of data to make relatively complex predictions and decisions.” Such algorithms can decipher text, segment issues, and anticipate how employees will respond to various initiatives. Moreover, machine learning can drive programmatic decision-making in a compliance program for employees through a “customer relationship management system”. The next step is what the authors term the “more sophisticated variant, deep learning, are the hottest technologies in AI and are rapidly becoming powerful tools in marketing.” That said, it’s important to clarify that existing machine-learning applications still just perform narrow tasks and need to be trained using voluminous amounts of data.
Stand-alone applications are “best understood as clearly demarcated, or isolated, AI programs.” Conversely, integrated applications are embedded within existing systems and such AI applications are often less visible than stand-alone ones. This allows employees to be delivered a more sophisticated solution for the Quote To Cash (QTC) or Procure To Pay (P2P) processes. With a stand-alone system, employees need to go to a dedicated app and request suggestions. It appears that compliance professionals will “see the greatest value by pursuing integrated machine-learning applications, though simple rule-based and task-automation systems can enhance highly structured processes and offer reasonable potential” for not simply more efficient compliance processes but for commercial returns.
For corporate compliance professionals with limited AI experience, perhaps the “way to begin is by building or buying simple rule-based applications.” You can start with “crawl-walk-run” approach. Once a compliance function acquires basic AI skills and an abundance of data, you can start moving from task automation to machine learning. Moreover, new sources of data, “such as internal transactions, outside suppliers, and even potential acquisitions”, are something compliance functions should have access to, since most AI applications, particularly machine learning, require vast amounts of high-quality data. Once again this is precisely what the Department of Justice (DOJ) specified in the 2020 Update to the Evaluation of Corporate Compliance Programs when it mandated that compliance have access to all corporate data even when siloed.
There are challenges in implementing an AI tool for communications as “implementing even the simplest AI applications can present difficulties. Stand-alone task-automation AI, despite its lower technical sophistication, can still be hard to configure for specific workflows and requires companies to acquire suitable AI skills.” It will also require “careful integration of human and machine tasks so that the AI augments people’s skills and isn’t deployed in ways that create problems.” The bottom line is that while AI holds enormous promise, for compliance professionals for a variety of uses, it still accomplishes only narrow tasks.
But it will be a journey for compliance. The compliance function “and the organizations that support it, IT in particular, will need to pay long-term attention to building AI capabilities and addressing any potential risks.” Yet compliance professionals cans start developing a strategy today to take advantage of AI’s current functionality and its likely future. Compliance communications to both internal and external stakeholders is certainly one use that should be on your horizon. When we receive the next iteration of the Evaluation of Corporate Compliance programs you may well see AI specifically called out as a tool, the DOJ may expect multi-national companies to have AI in place and be using for a variety of compliance activities.