The traditional "Beer Game" gets a makeover with the application of artificial intelligence.
A.I. and machine learning promise to revolutionize not just supply-chain education, but the everyday lives of planners and analysts, according to Sara Hoormann, principal with OpexAnalytics.
Q: Give us a quick reminder of what the Beer Game is.
Hoormann: Most people, especially if they've been in a supply-chain MBA or undergrad program, have probably sat down at a table and played the Beer Game. It’s a collaborative game where there are four spots on the supply chain. Everyone is looking to place orders to not run out, to make sure that they're supplying upstream and downstream. It teaches the Bullwhip Effect, caused by the innate, human panic when they receive demand that they know they can't fill. So they over-order, and then under-order, because they don't want to pay for overstock. It teaches you to look at the whole supply-chain, and the impact of visibility and information-sharing.
Q: The Beer Game has been an important part of supply-chain training for many years. Now we’re bringing artificial intelligence to it. How does that affect the use of this training tool?
Hoormann: We worked with a senior research associate at Lehigh University to show how A.I. and deep neural net activity could make operational decisions for a company. The idea was to train a reinforcement-learning algorithm to play the Beer Game, and show people what it's able to do. In the game, there are two options. You can play against the A.I., or you can play with it on your team, and see if it calms the panic and the Bullwhip Effect. In most cases, the A.I. wins.
Q: So the machine is making better decisions?
Hoormann: Yes, it’s finding patterns and teasing them out. That's what deep learning is doing. It’s able to go much further than people can go visually or manually, and identify patterns and features that lead to better decisions in the long run. But I would never say that complete automation is the way to go. If A.I. is making a better decision 90 percent of the time, then planners are then freed up to make more intelligent decisions about those other 5 or 10 or 5 percent. Then can just monitor it, instead of having to do all of the manual, laborious processes that they usually have to do. A.I. is going to find its niche in learning and making good decisions here for most of it.
Q: As the years go by, will machines take over more and more of these decisions, and the human role become more diminished?
Hoormann: Probably, yes. But it frees up humans to concentrate on things like acquiring new suppliers for a new product. Oftentimes when we talk about A.I., there’s this fear that it's going to replace everyone. It will replace certain job functions, but it will also allow humans to make the more crucial, one-off, alert-based decisions, as opposed to spending their time doing mundane tasks.
Q: Are we talking a technology that’s essentially reactive? Or is it something that has predictive capabilities, so that you can prepare for things in a better way?
Hoormann: It can do both. In the case of projects we've done with clients in the past, the system is just alerting the customer-service agent or planner to the potential for an issue based on something it's seen before, so they can take corrective action up front.
Q: How is this changing the skillset required of supply-chain professionals in the future?
Hoormann: We want people who are well-versed in all the technologies that are out there. But it’s also about how they approach a problem, and figure out the core drivers behind it.
Q: So we need more people with analytical skills?
Hoormann: Yes. It’s a challenge, though. We face this problem every day, when we’re looking at recruiting. We might have a person who’s super-technical, but also needs to be able to relate to the client, who knows the problem and application to the business really well. The team we have is very diverse; I think that's what it requires. It doesn't always come down to one person. It's a team environment, the bringing together of these two skills.
Timely, incisive articles delivered directly to your inbox.