Artificial Intelligence and the Job Market

Bill Cheetham has been a computer scientist at General Electric for the last twenty years. However, the nature of his work has evolved, as has the technology around him. Cheetham is now a part of GE's Industrial Artificial Intelligence Laboratory, a division of the conglomerate that helps solve problems using technology, including AI.

“I'm usually working on two projects at a time,” Cheetham said from his office near Albany, New York. “These can last anywhere from six months to three years.”

He described the process using one of his projects involving gas turbine engines that generate electricity for municipalities. His goal was to use AI to diagnose maintenance issues with the turbines, which were studded with 100 sensors that continually generate data.

The first step in the process was creating a team that involved Cheetham, another member of the AI group, four people from the gas turbine manufacturing company, and at least one of the end users from GE's information technology department.

To be more effective, Cheetham took classes in turbine maintenance and spoke at length with mechanics about diagnostic testing. Where it was once common to examine and interpret data by hand, pouring the figures into a software program such as Excel and then creating a graph based on the data, Cheetham was now tasked with implementing the use of AI technology to help maintain the engines. During this research process, he visited the manufacturing facility, saw the turbines in use, studied where the sensors were located, and noted the types of data they recorded.

After studying the system, Cheetham began determining how AI could be used to help solve turbine maintenance problems. Much discussion over the design for such a system occurred before any real programming was started. “That wouldn't even be the start of the code. We talk among ourselves to see how to improve [maintenance], then we go in and start coding, then we test the code, then go back to the users. If they like it, we'll expand on it. That's the discipline called software engineering,” he explained.

Cheetham's team wrote software with algorithms to sift through the voluminous data to detect flaws and even recommend repairs. His area of expertise is in case-based reasoning—reasoning through experience—using expertise from the past to solve a problem. In some cases, Cheetham said changes in manufacturing might be required that would then be discussed with the engineers. Similar work was done with diagnosing aircraft engines, and he said they pooled data from GE–built airline engines so the database was filled with information from thousands of engines, allowing the computer to find anomalies.