Can You Trust the Information Your Supply Chain System Tells You?

Can people really trust information produced from a system under high stress, high risk, supply chain disruption situations? This problem was indirectly explored by a group of researchers from NC State University’s Laboratory for Analytic Science. These researchers asked the following question: what happens when humans have to trust real-time data under the types of high-stress, high-pressure environments that are typical of major supply chain disruption events? Humans tend to rely on available information while completing complex tasks. But what happens when information is presented by human and automated sources? And what happens if those information sources conflict? This situation occurs more than you might think. For example, a Russian passenger jet and cargo plane in 2002 crashed in a mid-air collision. Automation told the two planes to change elevation in different directions, but so did an air traffic controller; these two messages directly conflicted in terms of the directions delivered to the pilots. One pilot listened to the automation and the other pilot listened to an air traffic controller, and the planes collided. In a series of tests with undergraduate students, the students were presented with the following scenario:

“You will be performing as the leader of a vehicle convoy. Your mission is to deliver critical supplies to a nearby warehouse. Your task will be to select a delivery route. Participants must select a route for their military convoy from three possible options.

You will be shown a map displaying three delivery routes. The map will identify the location(s) of past IEDs (Improvised Explosive Devices), as well as areas of insurgent activity. You will also receive information from a local intelligence officer who will provide you with additional data about the area.

Consider the three routes and select one. Make your decision as quickly as possible; you will have 60 seconds to complete this task.”

In this scenario:
• An automated tool provides a map that contains information regarding past IED explosions and insurgent activity to illustrate one optimal route choice.
• The human provides information that conflicts with the map and recommends a different route.

Under this simulation, the findings were very interesting:

• Presentation order did not significantly affect reliance/trust in human and automated sources within risky decision making tasks. In other words, there may be more critical design choices worth considering when designing systems to promote reliance for this type of scenario.
• When presented with conflicting information from automation and human sources in high workload scenarios, operators may increase trust in human sources. In other words, increased workload negatively affected trust in automation.
• When presented with conflicting information from automation and human sources in high risk scenarios, operators may decrease trust in automated sources. Increased risk positively affected trust in the human. This may be due to the added load of assessing automation’s trustworthiness.

The implication of this research is that any real-time supply chain system must have the following characteristics to be successful:

• The system itself must produce data that is trustworthy. In other words, the data must represent the reality of the situation.
• The information provided by the system must be aligned with human perceptions of what is happening as well. It also suggests that combining human observations with system data can augment and increase the level of trust that others observing the information will have in the data.
• Under high risk situations common in major disruptions, people will trust humans over system-produced data. This is a cultural artifact that should be considered, and an obstacle that may need to be directly addressed.
• Under a high level of workload stress, again operators may trust more in human dialogue. The need for human to human communication under these types of situations is important.

Because the emergence of real-time supply chain systems is so novel, there are many such cultural artifacts that may need to be overcome, even if the systems issues are addressed.

(Source: In Automation We Trust? Identifying Factors that Influence Trust and Reliance in Automated and Human Decision Aids. William A. Boettcher, Roger C. Mayer, Christopher B. Mayhorn, Joseph M. Simons-Rudolph, Sean M. Streck, Carl J. Pearson & Allaire K. Welk, Ph.D., Presentation made at LAS, March 23, 2016.)