Being Wise About Supply Chain AI

The shift will be gradual, as some tasks are better handled by AI than others, and job descriptions and skill sets will need to change.

Artificial intelligence is drastically changing the way the manufacturing supply chain works, from how human workers need to be trained to how managers oversee and run warehouses.

We can think of AI as a way of intervening between data processes and humans to offset humans’ lack of access or knowledge. Every part of the supply chain—sourcing and procurement, logistics, operations planning and control, and delivery, fulfillment and customer service—has numerous opportunities to benefit.

For example, AI can act as a project manager of sorts between suppliers and customers. AI has the ability to coordinate planning and delivery between manufacturers and suppliers, allowing manufacturers to respond to suppliers’ changing needs without the inefficiency of human error. Specifically, machine learning algorithms can incorporate past decisions on schedules, forecasts and actual deliveries, relate those to subsequent financial performance, and adapt decisions to maximize both revenues and profitability. AI will explore every potential combination of demand, product mix and delivery times to develop more robust and accurate forecasts. As for customer relations, chatbots are already available to address customer service issues and solve problems without the need for human workers. Increasingly, when customers log on to manufacturing suppliers’ websites, they see a chatbot option pop up.

Logistically, the potential around autonomous vehicles and autonomous delivery is the next big frontier, and there’s a lot of excitement in the possibilities. For example, the Volkswagen Transparent Factory in Dresden has autonomous vehicles that use AI to move Volkswagen parts around the plant floor. The ability to have almost fully automated machining and assembly of parts in manufacturing already exists.

Job Changes Afoot

While the implementation of AI can improve efficiency and productivity within the manufacturing sector, the reality is that organizations aren’t going to adopt this new technology overnight. And as the use of AI becomes more prevalent, it will become increasingly important for organizations to ensure that employees are properly trained to use the technology. We’re using AI to alter and augment the manager’s role, meaning that project managers will need a much deeper understanding of computers and computer programming. Being expertly versed in data manipulation within the AI system and specifically the scope and limitations of such systems will be a core part of the project manager’s role and other operational manager roles.

Human project managers will also have to fundamentally understand the supply chain in order to strategically correct problems when things go wrong. Humans’ innate capability to handle multi-dimensional problems (e.g. balancing options for promotion policies, team configuration or motivation with daily planning and control) will continue to be necessary when schedules change, demand shifts, a natural disaster occurs or machines break down.

Project managers working in manufacturing will need to understand how to properly fight the fires that arise when these systems break down, understanding that without the intervention of human problem-solving abilities when there’s a problem that can’t be solved with AI and algorithms, the use of AI can harm the supply chain.

Humans have a significant edge over machine learning in AI: they can handle complex trade-offs between options and be innovative by drawing on knowledge, experience and capabilities that are broader than the typical AI algorithm. AI is best at performing specific, clearly defined tasks, such as warehouse picking; it is less effective at more nuanced activities, such as developing promotion strategies and supply chain innovations.

Implementing AI in the supply chain effectively will take time and planning, and if manufacturers do not take the time to do this properly, they will be more likely to find themselves in a “garbage in, garbage out” situation. Take the classic example of Microsoft’s chatbot on Twitter: In less than 24 hours, the conversation with the chatbot disintegrated into bigoted, racist, misogynist, and problematic behaviors because the machine interpreted a biased stream of data from the Twitter feeds it was programmed to learn from. A chatbot’s ability to sense nuance is limited and leaves opportunity to miss many of the variables that humans are able to sense. AI needs to be backed up with a comprehensive understanding of how supply chains work, and frankly, I don’t think many organizations have those systems in place yet.

AI limitations can cause problems that disrupt supply chains and customer service systems, potentially jeopardizing entire manufacturing operations. I have witnessed new manufacturing technology mishandle entire operations because the managers weren’t aware of the system’s capability or constraints. I worked for an engineering company in the United Kingdom that was among the first to introduce CNC manufacturing equipment (computer controlled machines), and it bankrupted the company. In theory, the company had increased its capacity exponentially; however, in practice, its use had almost disappeared because every single job needed programming, and the programming time was astronomical. The company couldn’t keep up with the programming needs.

Augmenting systems with the use of AI is where the opportunities lie, not through fully replacing humans with artificial intelligence. For a production planner, for example, having AI systems that provide greater visualization of data relating to complex supply chain processes can offer a much broader picture. Using AI can help overcome many of the natural perceptual biases humans have, such as recency bias, groupthink, and those resulting from an individual’s lack of experience. AI-based systems will not only present data in easy to understand formats (such as charts and graphs), but even more significantly, they will shape how we interpret such data by suggesting solutions we might have never considered on our own.

While AI usage can increase efficiency exponentially in the manufacturing sector, there are many pitfalls and potential problems if the technology is implemented before it’s fully understood and before human workers are properly educated and trained to use it. In short, a systematic and careful approach to incorporating AI into the manufacturing industry could prevent catastrophic issues and system-wide failures while increasing production capabilities in the long term.

Simon Croom, professor of supply chain management, joined the University of San Diego School of Business in 2005 and was appointed executive director of Supply Chain Management Institute and academic director for the Master's in Supply Chain Management program. In this capacity, he oversaw the growth of supply chain management in the school, including the attainment of the first national rankings and recognition for the supply chain programs. 

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.