Among the many benefits that artificial intelligence can bring to industry is to reduce risk in every link of the supply chain, says Jim Hayden, chief data scientist at Everstream Analytics.

Artificial intelligence might not factor into every link in the supply chain, but it’s certainly playing a role in solving risk-related problems, Hayden says.

If anything, AI is good at generating data, which can be used to build training sets for models to learn from, he says. That’s because AI is good at summarizing lots of information. “It can help you get to the salient things going on in supply chain without you having to understand all that information.”

Hayden cautions, however, that the content generated should be considered as a “first draft.” He and his colleagues use the data to understand a risk event, but not before human analysts “refine it and make sure there's no ‘hallucinations’ in what’s generated.”

It's clear, then, that AI isn’t plug-and-play. Hayden notes that the technology is probability-based. “You need to differentiate that from engineering, which is a little more rule-based black-and-white, when you're making decisions.”

He says the process takes more than just training a model. “You've got to build the pipeline to give data to the model, then test and retest it and make it better. Then you've got to deploy it into a production environment. All of that takes slightly different skill sets: engineers, data engineers, dev-ops people. So there's an entire cast of characters required to be able to deploy it.”

From a risk standpoint, AI can inform about dangerous events sooner. “It's going to be relevant risk, and that's what's key to making decisions. Only tell me what I need to know, make sure it's relevant, and give me enough context so I can understand the potential impact to my business.”

For information on the next Gartner Supply Chain Symposium/Xpo, click here.