Artificial Intelligence. Machine Learning. Predictive Logistics.
These topics are getting a lot of attention today, but they’re also generating a lot of questions from supply chain and logistics executives. What are these technologies? How can they help? What’s real and what’s hype today? What should I do to get started?
Those are some of the questions I discussed with Adam Compain, CEO at ClearMetal, in a recent episode of Talking Logistics. I kicked off the conversation by asking Adam to provide a clear definition of these terms and to share his perspective on why there’s growing interest in these areas. Here’s a bit of what of he said:
The most important thing to understand is that when we talk about AI and machine learning what we’re really talking about is a fundamentally different type of computing than has existed before. If you want to simplify it, the analogy I often use is that it’s the difference between a smartphone and a flip phone…It is a type of computing technology that is a necessary co-pilot to making the very complex decisions we make every day, whether it’s personally using a smartphone to navigate the world in traffic, or in the supply chain to navigate all of the exceptions and contingencies that go awry when you’re dealing on a global scale.
If you look at the business leaders out there — like Amazon, Facebook, and my former employer Google — they have all said that everything they do will be AI assisted from now on…It’s very much like the wave we saw when business went from analog to using computers to do things differently and make better decisions.
Getting to the definitions, when we speak of Artificial Intelligence, it’s really an umbrella term for a set of technologies that leverage this different kind of computing. It’s named AI because, in many ways, it mimics the way humans behave and think and pattern recognize; it just does it at a level that is many orders of magnitude more sophisticated.
When we talk of Machine Learning, one of the core tenets of AI, it is a different kind of computing in two ways: first, machine learning means that instead of programming and prescribing a static algorithm to understand the way the world works — putting data in one side of the algorithm and getting an output out the other side like an equation — machine learning instead looks at a ton of data and from it discovers and recognizes patterns that would otherwise be unforeseen to the human eye or to the best-trained statistician or person who writes algorithms.
Secondly, what machine learning does, is that every new piece of information or data it receives, it actually learns over time. This mimics the ways humans do it, but in both cases, it can see patterns and correlations far better than we can, with a level of perception that is far beyond what humans can do.
“Historically, what we’ve seen is supply chain operators focusing largely on physical operations and physical economies of scale to solve problems,” added Compain. “But coming from a place like Google, what we learned is that there are ways to leverage technology and be smarter as opposed to being bigger to solve problems.”
And that’s where Predictive Logistics comes in. “It’s about using intelligence over scale,” says Compain. “It’s about being proactive and predictive instead of reactive and just managing exceptions after the fact. It’s also this notion of companies wanting to be much more marketing-driven, customer-driven, and data-driven rather than operationally based, so this category is [enabling] the next generation of supply chain and logistics to get there.”
Compain also talked about the importance of data in enabling predictive logistics, along with a common misconception:
Data is obviously needed to get to the future of predictive logistics and enhance predictive visibility, but there’s a misconception that you need different and new kinds of data, like needing all containers or pallets to be equipped with sensors before you can do predictive logistics. That’s actually a fallacy. There’s enough data in the industry that you can leverage, as long as you can make sense of it using machine learning, to do things like predictions and obtain intelligent insights using the data you already have.
Compain shared more on this topic in a recent guest commentary, The Takeaway from CSCMP Conference: The Data Problem Runs Deep.
I encourage you to watch the rest of my conversation with Adam for additional insights and advice on this topic, including some use cases for applying predictive logistics today. Then post a question or comment and keep the conversation going!
