Artificial intelligence (AI) can help organizations save money on supply chain costs and save time on employees' most repetitive tasks, but before they can apply AI, companies must first learn to trust the data that underlies the technology, according to a panel discussion held Tuesday at the CSCMP EDGE 2017 annual meeting in Atlanta.
A lot of companies already have the data they need to apply AI, but they're not using it well, and they're not trusting the results, said Abel Muniz, regional operations manager for the Americas at Panalpina Managed Solutions, during a session titled "The Artificial Intelligence-Based Supply Chain." For example, humans often pad estimates about transit times so they won't get blamed if a shipment is delayed. "Data doesn't have an ego or a bias," Muniz said. "So if the data says the last thousand shipments took 25 days," he continued, his company would base its decisions on that accurate date, not on a biased estimate. Data-based analysis can reveal surprising facts that had previously been obscured, Muniz said. "The biggest 'aha!' moment for a lot of our customers is seeing how fat they are on shipping estimates and inventory carrying costs."
Three ways to convince a company to trust the results of its AI project are to: tie the project to solving a specific problem at the firm, educate colleagues that AI is not the "Star Wars" product many people imagine, and prove that the project can produce a profitable business value, said Alejandra Dorronsoro, senior international logistics manager for GP Cellulose, a unit of Georgia Pacific.
Dorronsoro helped apply an AI system to a persistent problem at GP Cellulose, using the technology to create more accurate and reliable predictions of delivery dates for shipments to customers. The application also helped cut the time that employees had spent organizing and cleaning up databases, freeing them to focus on more creative tasks like developing strategies, managing relationships, and being innovative, she said.