Supply Chain Discussion: A.I.: Buzz vs. Reality

In this Discussion, Robert Bowman from Supply Chain Brain speaks with Robyn Winner, head of demand generation with Loop. Here she relates some of the misconceptions that surround the application of artificial intelligence to the supply chain — as well as the real benefits that the technology can deliver.

The biggest misconception around AI today is that “in and of itself, it is the end result,” Winner says. She likens that belief to the dawn of the internet, when users weren’t clear on what to use it for. AI, she adds, is “a means to an end.”

That said, it’s important to understand AI’s limitations. Generative AI, as a large language model or LLM, is prone to “hallucinations” if it hasn’t been properly trained on the information being sought in a query. It’s fine for asking questions that solicit an opinion, Winner says, but not so much for those requiring “binary” answers. “It’s not a magic solution.”

AI can be of significant value when deployed to solve specific problems, and automating certain human processes such as pattern recognition. The model can be queried, for example, about where the most inefficient parts of the business are. It also does a good job of removing the need for physical processing of paperwork.

One of the best applications of an LLM is supporting a customer success team, providing answers to questions posted live on the phone. Tougher to implement, but also beneficial, is tapping into a model’s deeper intelligence layer for making sense of large amounts of data. In such cases, AI can build a digital twin that allows users to run multiple “what-if” scenarios for key aspects of decision-making — although in most cases, the choice of the best option continues to rest with humans.

“At the end of the day, we as people should be there to make decisions.” - Robyn Winner

Source: Supply Chain Brain