Generative AI in operations management

The state of artificial intelligence (AI)

The progression of technology is enabling a future where everyone, regardless of their technical background, will be able to communicate and work directly with machines to automate a range of tasks. The recent unveiling of Chat Generative Pre-trained Transformer (ChatGPT), developed by OpenAI, has shown how far we’ve come. Put another way, the platform has demonstrated how powerful AI can be in removing the barriers between humans and machines.

According to recent research, leveraging the US Census Bureau’s 2019 Annual Business Survey, firms that adopted specialized software, AI, or robotics witnessed 11.4% higher labor productivity.¹ However, the firm adoption of AI and robotics was still limited in the United States—the share of firms that adopted these technologies in 2018 was 3.2% and 2%, respectively, while the share of firms that adopted specialized types of software such as enterprise resource planning (ERP) or warehouse management system (WMS) stood at 40.2%.

Undoubtedly, we can expect the share of AI usage among organizations to grow exponentially over the coming decade. According to Gartner, in 2022, CEOs cited AI as a top priority for the third year running.² But on the back of recent technological leaps in generative AI, the question is how the technology will be utilized to enhance productivity within firms and how it will affect workforce planning for executive teams.

Five ways generative AI will affect fulfillment operations

Focusing on fulfillment operations, the introduction of technologies such as WMS and transportation management systems (TMS) enabled firms to increase labor productivity by automating and simplifying processes. Moreover, these technologies increased the relative demand for more skilled workers in fulfillment operations. For example, the introduction of WMS required firms to hire or outsource WMS administrators to configure systems and test new software enhancements and improvements. At the same time, the introduction of automation equipment reduced demand for lower-skilled tasks such as putting products away or loading a truck.

With the rise of generative AI alongside the proliferation of low-code or no-code platforms (essentially more user-friendly software platforms that can be used by non-programmers), we expect the composition of labor demand to shift even more toward skilled workers.³ What’s more, we expect the line between operations and technical staff to blur as advancements in AI make technical changes more accessible to operations staff. Below, we outline five areas where we believe generative AI will have the most significant impact on the fulfillment workforce and operations.

1. Training: Because generative AI allows users to query unstructured data sets and receive structured and organized responses, we envision a shift toward more user-led training for WMS, TMS, or operational enhancements—one where a new user can essentially “have a conversation” with the system (or repository of training content) to learn functionality and best practices.
2. Software trial and error: As generative AI learns a particular system, we see a world where a user can operate and test software enhancements—such as changes to picking or route management—without having to bring in software developers.
3. Root-cause analysis: Such a system will enable operations teams to perform a root-cause analysis in real time to understand where the system is having difficulties or not meeting their expectations. For example, one will be able to directly ask the system why a particular product isn’t in the correct location instead of getting an error that it isn’t there.
4. Reporting: In the future, instead of enlisting business intelligence (BI) developers to create or modify operational reports, an operations or floor manager would simply tell the system to auto-generate any desired reports, such as the order fill rate on January 2 or the number of put-away transactions performed by Juan or Sally in the first six months of the year.
5. Communication and collaboration: Finally, and perhaps most importantly, generative AI will enable fulfillment teams to structure, organize, and streamline disparate data sets and knowledge across the network. This would enhance communication and collaboration to enable cost optimization and a superior customer experience. This can be anything from quickly summarizing Zoom calls to customizing slides and visuals for a specific audience.

In short, generative AI will enable operations teams to make smarter, faster decisions—enabling continuous improvements throughout the fulfillment network. With generative AI, the system becomes both the monitor and investigator, allowing managers and staff to focus on decision-making rather than troubleshooting root causes.

Preparing for the future

These impact areas are by no means comprehensive. As with the dawn of the internet, we simply can’t predict with any degree of accuracy what jobs will be created, destroyed, or augmented by generative AI over the coming years and decades. However, we can say that AI will have a monumental impact on the future workforce and, in turn, on how firms approach workforce planning in the short and medium terms.

What will this look like in the world of supply chain distribution? Operational leaders will increasingly need to prioritize problem-solving and analytical skills as core skills for all staff—from forklift drivers to site leaders. At a higher level, it’ll mean building a lean operations workforce that 1) deeply understands the operation, 2) is aware of real-time changes in the network, and 3) can manage and solve operational problems, with the help of generative AI, to fulfill customer needs and reduce operational costs.

Authors:

Endnotes:
¹ Darren Acemoglu et al., Automation and the workforce: A firm-level view from the 2019 Annual Business Survey, NBER Working Paper No. 30659, November 2022.
² Gartner, “Gartner survey finds CEOs cite AI as the top disruptive technology impacting industries,” press release, May 17, 2023.
³ Craig S. Smith, “‘No-code’ brings the power of A.I. to the masses,” New York Times, updated April 1, 2022.