The increased use of technology will continue to drive the performance of corporate compliance programs. Chief Compliance Officers (CCOs) and compliance practitioners must understand how to incorporate these technological innovations into their compliance program from the legal perspective, as continued evolution will be required by enforcement authorities, such as the Department of Justice (DOJ) and Securities and Exchange Commission (SEC), under anti-corruption laws such as the Foreign Corrupt Practices Act (FCPA). However, an even more important driver will be the continued business response demand that compliance brings not only more efficiency, demonstrates a return on investment (ROI) but also benefits to the bottom line of any organization. Over the next couple of blog posts, I will be exploring the uses of Artificial Intelligence (AI) in a best practices compliance program and its collaboration with the compliance professional. Today, I consider what a compliance professional can bring to an AI solution.

In a Harvard Business Review (HBR) article, entitled “Collaborative Intelligence: Human and AI Are Joining Forces”, H. James Wilson and Paul R. Daugherty explore how companies are seeing and using not simply the intersection of AI and business but also the collaboration of AI and business to “reimagine their business processes, focusing on using AI to achieve more operational flexibility or speed, greater scale, better decision making, or increased personalization of products and services.” That last sentence sounds like a good goal for every compliance program to strive for going forward.

The authors found there are some basic principles that underscore the assistance companies receive from the collaboration of AI and their workforce. They include the ability to reimage their business process; more fully grasp the employee involvement in this experimentation; use employees to provide more focus on AI strategy, the responsible collection of data, to redesign work processes to incorporate AI into them and to cultivate related employee skills. The authors found that a survey of 1075 companies in 12 industries, “the more of these principles which were adopted, the better the AI initiatives performed in terms of speed, cost savings, revenues and other operational measures.” Obviously, AI and humans have different strengths and weaknesses. Humans have greater leadership, teamwork, creativity and social skills (at least one would hope so). AI brings greater speed, quantitative and scalability to the equation.

Training AI

The authors believe that humans bring three important roles to the AI/human collaboration arena. The first is around training. It all starts with massive amounts of data. Yet not all of that data is in forms that can be immediately machine learnable. Consider the anti-corruption world and the number of names there are for bribery schemes and how employees try to hide such communications. Further, consider the number of mechanisms employees use to hide and disguise bribe payments inside ERP systems.

Next, consider the merging issue of bots being used to help facilitate compliance communications. The authors pointed to the Microsoft AI assistance Cortana, which “required extensive training to develop just the right personality: confident, caring, and helpful but not bossy.” The same can be said for almost any compliance professional so it should not be surprising that the same is required for a compliance bot responding to inquiries from employees or when a bot is employed to help navigate an employee through internal reporting mechanism and process. This can mean steps beyond such basic inquiries as “Sorry to hear that” or “Can you be more specific?” but even to channel the emotions into a positive action.

Explaining AI Decisions

Here the authors speak to the ‘black-box problem’, which is where human experts are required to explain how AI reached its conclusion. Such explainers are critical in evidence-based professions such as compliance “where a [compliance] practitioner need to understand how an AI weighs inputs”. The authors believe that such explainers are becoming not simply mandatory but the norm in regulated industries. However, that regulation is far beyond banks, financial institutions, insurers, medical and pharmaceutical companies. Consider the requirements under General Data Protection Regulation (GDPR) that companies must give consumers an explanation for an algorithm-based decision such as a credit score or mortgage rate.

Sustaining Decisions

There is another function for humans, which is to make sure AI systems are “functioning properly, safely, and responsibly.” This is not simply the Trolley Car hypothetical or even self-driving cars but moves towards ethical issues in AI decision-making. If AI decisions are found to have a disparate impact on one group or area, it could signify illegal discrimination. In such a situation, ethics managers will be responsible for detecting, preventing and remediating the issue. Similarly, these ethics managers will be required to make sure the plethora of internal policies and procedures are followed as well as such regulations as GDPR noted above.

Nevertheless, this sustainment will go beyond making sure there is no discrimination and the rules are followed. AI must use and manage the information responsibly. Facebook cannot seem to either get its story straight about how its users’ data is sold/used/given away to third parties for whatever use the third parties want to make of the data. Facebook has certainly not moved in any meaningful way to manage these third parties through oversight, monitoring or auditing. This means more human levels of prevention, detection and remediation, which are the three most components of a compliance program and which every compliance practitioner engages in their day-to-day practice of compliance.

Tomorrow we will consider what AI can bring to the compliance practitioner.

[View source.]