What is the most important decision made by your company? A decision so important that the profitability and reputation of your company rely on getting this decision consistently right? Who, in your company currently makes this judgment call? If they are the best at making an important decision we call them experts. Do you have the expert who makes that critical decision in your mind right now? What would happen if your expert left the company unexpectedly for a new job or retirement?
What if I told you there is now a way to duplicate that decision made by your expert and make it possible to scale that decision so that others in the organization could share that expertise as well? That is, a technology that makes expertise available to everyone in your organization who could benefit from it? How many dollars are you spending on traditional training methods to raise the game of others that don’t really work? If all your employees had ready access to the best of the best decision maker how would it impact the bottom line of your company?
A new decision-making technology that makes this possible is called TOM, a Tacit Object Modeler. TOM is an artificial intelligence decision modeller that enables the creation of virtual experts. TOM models the judgment, intuition, and years of experience of experts in making time-critical, high-consequence subjective decisions. Virtual experts are software models that replicate the decision-making ability of a human expert in a specific domain. TOM eliminates “noisy” data problems to arrive at an optimal solution – an expert’s specific decision-making behavior. TOM also addresses a primary challenge that organizations face: how to close the gap between their own standout performers (experts) and novices, less effective decision makers. Many people are reluctant to use virtual experts to make superior judgments because they don’t understand or trust the underlying judgments of human experts. This is where TOM serves as a consistent and impartial tool that produces an algorithm to replicate and document the decision-making processes of specific experts.
It has long been a challenge for humans to model the decision-making process because we lack the clear and robust data needed to “mine” for the result from every possible situation and potential outcome as decision-making environments change rapidly in short time frames. This is why traditional artificial intelligence using “big data” models have not been particularly effective when tasked to replicate subjective human judgment—it is impractical to ascertain decision processes by asking the experts their judgments across every possible situation. The main problem is the more data you get, the less you know and while big data is important it’s the “small data” that is more actionable. In addition, decision data sets themselves are noisy, have limited shelf life, and there is no efficient way to know what data to include as a foundation in these cases if the objective is to duplicate human thinking.
TOM minimizes such limitations by interacting with humans directly. TOM does not analyze a situation per se but instead decodes how experts execute their subjective judgment across the spectrum of experiences. Thus, there is no need, to begin with, a large data set. While traditional AI approaches are essentially advanced data mining exercises, TOM mines human expertise by starting with the decision itself.
The difference between mining data vs. human expertise is the difference between explicit knowledge, which is known universally available to others, and tacit knowledge which is subjective and what the expert uniquely brings to the table. We all know executives who can see around corners, they just know what lies ahead and are usually correct. It’s that tacit knowledge that sets the expert and the non-experts apart that TOM taps.
After an expert’s decision variables are established, an advanced sampling algorithm and rule extraction enable a reproduction of the expert’s decision-making algorithm. The algorithm then reliably predicts the expert’s decision given any set of his or her variable preferences. TOM learns to replicate the decisions the expert would make across many different, as well as inexperienced, scenarios.
TOM has been successfully implemented across a number of industries including, healthcare, financial services, transportation, human capital, strategy, marketing, and sports. TOM is applicable in any context where an expert is making an impactful subjective decision. Here are just a few of the important expert decisions that TOM has successfully duplicated:
As a control center engineer, given the current conditions and stresses on the power grid, what actions, when and where, are best to keep my country's power grid stable?
- As a foreign exchange trader, given the current factors of the market, exchange rates, custom segments, product, transaction size, agreement type, what margin should I assign?
- As a parole officer, does this parolee require immediate intervention?
- As an anti-money laundering/terrorist financing investigator, do I escalate this alert now?
- As a human resource officer, is it time to intervene regarding this employee’s absenteeism?
Using TOM in the cases above has resulted in greater productivity and cost savings for the organization in large part to the reduced performance gap between experts and non-experts. As the future unfolds, decision-making AI will radically increase productivity across the economy by elevating the skill levels of the employees who learn from experts’ decision-making. The challenge will be preparing organizations and their leaders for the exponential economic growth that will follow this decisive competitive advantage.
Karl Kuhnert and Thomas Mark Keith, Managing Director, ai Innovation Technology, LLC. email@example.com