As artificial intelligence (AI) continues to advance, there seems to be a growing fear that AI systems will replace humans in every conceivable professional field. While that is true in some cases (e.g. taxis will one day be driverless), humans for the most part will shift into new roles that accentuate the “heavy lifting” AI will begin to do for us.
A great example is when companies implement an automated auditing and compliance monitoring system like Oversight Insights On Demand to take over much of the manual work that human auditors have done for decades. The human auditors are not being replaced; they’re simply offloading tedious data crunching to a system that will bring them issues requiring human decision making.
That said, an essential characteristic of an AI system is its ability to “learn” from the humans who resolve the problems the system flags.
For example, hotel charges from the CEO may be flagged because they’re higher than usual, or employees may get dinged for putting alcohol on expense reports.
What the automated travel and expense auditing system doesn’t know is that the CEO has higher travel expenses because he travels with an entourage and puts them on his hotel charges, or that employees are allowed to expense alcohol when they’re entertaining clients.
These exceptions are opportunities for AI to learn by one of two means:
- By independently tracking and understanding what humans do with the findings it brings them (called implicit machine learning)
- By humans guiding the machine learning directly
AI systems improve by learning from us, which is why it’s important to have resolution automation tied to the analysis – it creates data for the AI system to learn from.
The biggest challenge companies face when implementing an automated auditing system is staying focused on using the data generated by the solution, not just having it.
Analytics are nice, but they’re not the end goal. In fact, companies that make the data the be-all-end-all are missing the greatest opportunity that AI affords them: the chance to continually improve their processes through the magic of machine learning.
The Importance of Continuous Improvement
Companies must always seek to improve their processes. In manufacturing quality improvement, there is an expression: “lower the water to expose the rocks.” The idea is that you want to remove “work in process inventory” to expose the places where the manufacturing process could flow more smoothly.
By lowering the water (work-in-process inventory), the biggest rocks (obstacles) are revealed so you can identify them and plan how to remove them.
You tackle the biggest obstacles first since you can’t fix everything at once. This process continues step-by-step until all the obstacles are gone.
When it comes to an automated auditing system, lowering the water is applying new analytics that let you attack new issues. Address those issues, and you lower the water a bit more. You use metrics to drive process improvements, so you lower the water again.
You just keep repeating that process, and you keep improving. With an AI system that is constantly learning and improving, there are always opportunities to get better.
No process is ever perfect because business is always changing. Vendors and employees come and go. People using the system will change. Some will be less willing to learn, while some will forget what they learned. New problems will always arise.
Opportunities for Machine Learning
But with new problems come new opportunities for learning. This is again where humans are essential for AI to function at its peak potential.
You may learn something working with one customer that suddenly makes sense with other customers, so you roll that process out with others. That is not a decision the AI could have made, but since you made it, your system is learning from you.
Having an AI system usually encourages a growth mindset for all employees.
It’s rare for an organization to have the capacity to introduce the automated auditing system to everything all at once, so employees must be asking themselves, “Where the next place we can optimize our performance by implementing the robo-auditing system?”
You may begin by applying it somewhere straightforward, like travel and entertainment, but as you gain experience you can focus it on accounts payable, accounts receivable, inventory management, contract labor, or contingent labor.
AI is a blade that is constantly sharpening itself, but it must be wielded by humans who are willing and able to use it to build a more efficient and profitable company.
This is part of an ongoing blog series based on the recently released book, “Robo Auditing: Using Artificial Intelligence to Optimize Corporate Finance Processes” by Patrick J.D. Taylor, Manish Singh and Nathanael L’Heureux.