Machine intelligence and human creativity in supply chain planning

addtoany linkedin
Robot touching information that is displayed floating in the air

I am reading this absolutely fascinating book “Deep Thinking: Where machine intelligence ends and human creativity begins” by Garry Kasparov, former world chess champion.

As the title suggests, in this book Kasparov shares a highly provocative point of view on artificial intelligence and its implications for the human race, with the backdrop of his 1997 loss in a highly publicized chess match up against IBM’s chess computer Deep Blue.

The book did make me reflect on my own experiences and views on the division of labor between the machines and human supply chain planners. Much has been written and said about how machine intelligence is impacting supply chain planning in the form of automating a human planner’s function, with implications on the future of the profession itself.

I would be remiss in stating that automation will have no impact on planning profession. Yes! The focus on automation in planning is increasing and will continue to increase. However, this has to take place in the context of empowering planners and significantly augmenting their productivity to handle activities with larger scope and with higher levels of cognition that can drive strategic value for business. When done in a thoughtful and deliberate manner, automation initiatives can significantly benefit planners who are willing to adapt and change, and organizations as a whole.

I come across many organizations that made extensive investments in advanced planning technologies with the intent of bringing more automation and standardization into their planning process. However, the tragic reality is that most of these deployments languish, only to see the planners bypass these systems to revert back to their excel spreadsheets and manual means of planning. Recently, I met with a large global process manufacturing company, which after a multi-year deployment of an advanced planning software ended up with 800+ excel spreadsheets that are used to manually generate and adjust base plans and run some basic simulations. The reasons for failure of such automation initiatives are three-fold:

1. Automation of “As Is” processes: Technology deployments should provide a great opportunity to rethink the as-is processes in light of the emerging best practices and the evolution of machine intelligence. Newer technologies provide tremendous potential to move to an entirely new way for planners to co-operate and collaborate with their peers to enable a digital supply chain. A very senior executive may buy into such a truly transformative vision. However, by the time the project gets into the design phase, unless the executive who bought into the vision is very deeply engaged, it ends up devolving into an expensive redo of the “as is” processes. This leaves very little incentive for the planners to adapt to the new system, resulting in them falling back into the all too familiar ways.

2. Lack of plan explainability and transparency: Let us face it! Supply chains and planning for them are far more complex than ever. In most companies, the bulk of a planner’s time is spent on generating base plans and synchronizing demand plan to supply plan to capacity plan to production schedules. However, most of these planning processes happen in batches with different cadences and frequencies. Most advanced planning commercial software packages subscribe to such batch oriented paradigm only to reinforce it. Keeping these plans in sync as the data constantly changes is quite herculean and results in a high degree of planner frustration.

This problem is compounded by black box solvers which don’t provide transparency into how plans are derived. If a planner does not understand why a particular order is getting shorted or why a capacity is getting overloaded, she doesn’t feel confident releasing the plan to execute. Instead, the planner spends an inordinate amount of time trying to validate and make sense of the plan. This proves antithetical to any automation efforts in planning and significantly increases manual activity due to increased time spent on plan validation. Most plans in such situations are dead on arrival by the time they are released for execution. Eventually, planners step outside of their painstakingly deployed commercial applications and resort to excel spreadsheets, defeating any automation efforts.

3. Inability to simulate a range of possibilities: As supply chains become increasingly volatile due to demand and supply side events (positive or negative), planners need to simulate a variety of “what if’s” to have multiple plays ready as the situation calls for. They need to do this tradeoff analysis in near real-time. However, the aforementioned batch oriented systems are latency ridden and do not provide quick simulation capabilities. Hence planners resort to downloading the planning system of record data into excel sheets and running simulations (with all too simplistic assumptions) so they can wrestle back control from the black box, batch oriented solvers, and in the process, defeating the automation efforts.

In light of the aforementioned challenges, some progressive companies are adapting a “concurrent planning” paradigm wherein they are, in near real-time, simultaneously planning across demand, supply, inventory, and capacities in a fully automated manner. This makes end-to-end plan synchronization a natural outcome, freeing up planners’ time for higher value add activities.

A concurrent planning system will have network-wide visibility, along with business rules such as demand priorities, preferred supply paths, lead time and capacity constraints, build ahead limits, expiration policies, and such, all of which are automatically factored into the plans. An “always on” near real-time planning capability eliminates latency in processing data into information.

With automated “concurrent planning” as the core foundation, planners bring in the art form by creating scenarios to support the increasingly dynamic nature of their supply chains, truly enabling business value. Running the scenarios and generating scores is done by the machines in near real-time.

Planners can now share these scenarios with their colleagues within the enterprise or with external partners across their network to drive system wide benefits, thus bringing in the best of human-machine collaboration. All in all, I see automation doing the heavy lifting for repeatable and necessary processes so the human planners spend time in driving higher value business decisions.

Machine intelligence has not evolved to the extent of perfecting the “art of supply chain planning” driven by the human intuition, the contextual knowledge of the business, and the ability to connect the dots between seemingly unrelated pieces of information from multiple domains.

Investing in planning automation is a perfectly valid strategy as long as organizations are deliberate in thinking through what activities are best suited for machine intelligence and which are best left to humans. The key here is to enable “Augmented Intelligence” that brings together the best of humans and machines! In the aforementioned book, Garry Kasparov cites an interesting fact.

The 1990 rating list of top one hundred chess players in the world included over twenty active players born before 1950. By 1995, there were only seven players of this group remaining in the top one hundred. Kasparov attributed the difference to the democratizing impacts technology had on the chess world.

The older players who employed human assistants to practice their game no longer had distinct advantage with the rise of computer chess. The players who embraced computer chess, regardless of their origins and age were able to sharpen their game. The players who clung to the old ways faded away.

Similarly, supply chain planners who stay abreast of the emerging processes and technologies, develop end-to-end concurrent planning thinking and make continuing education a habit will thrive. The individuals who rest on their laurels may find themselves increasingly irrelevant!

Leave a Reply