Despite advances, developing workplace systems that facilitate AI-human working relationships is still in its early stages. Companies need to invest in systems that enable robots and humans to assume more sophisticated roles gradually.
By Ken Cottrill, editorial director, MIT Center for Transportation & Logistics
As AI advances in the workplace, more applications that augment, not necessarily replace, human capabilities are emerging. It is essential to understand how these technologies can be integrated into collaborative partnerships with employees as we prepare workforces for AI-related transformative change.
Different combinations
In his book The Magic Conveyor Belt: Supply Chains, A.I., and the Future of Work (MIT CTL Media, 2023), MIT CTL director Yossi Sheffi describes the roles humans and machines play in controlled activities.
“At one extreme, a person might be fully in the loop, in that they must execute one or more essential steps every time the task must be done,” writes Sheffi. Automation handles certain aspects of the task, but human workers remain an indispensable part of the process.
Another workplace configuration involves a machine automatically processing most of the routine elements of a task but alerting humans when exceptional, anomalous, or complex issues need to be addressed. The task is automated most of the time. A more advanced version of this arrangement involves a person monitoring the task through a dashboard, intervening only when an alert is triggered.
In some cases, human involvement occurs only at a higher level, like designing a fully autonomous, machine-driven system. The system operates continuously without requiring human participation. An engineer might design a warehouse refrigeration control system that operates autonomously, eliminating the need for workers or managers to adjust the system.
Bot-powered procurement
Utilizing AI to augment human capabilities can free employees to focus on strategic challenges and those that deliver high returns.
For example, a leading retailer is using AI-driven bots in its procurement function to negotiate low-value, often infrequent, purchases from suppliers where the potential returns for the company are relatively modest. The retailer described the arrangement at MIT CTL’s Crossroads 2025 conference on March 17, 2025.
High-value negotiations with strategic suppliers are still overseen by human procurement managers. Deploying bots in this way enables the company
to utilize its in-house procurement expertise more effectively.
The bots also provide valuable intelligence that the company is using to improve its procurement operations, explained the retailer. This includes identifying which days of the week tend to yield the best outcomes, as well as more precise information on what constitutes successful negotiations. Interestingly, the retailer has found that including bots in sourcing events can remove the emotion from interactions with suppliers and promote smoother negotiations.
Warehouse partnerships
In supply chains, a useful example of a place where different combinations of human and machine capabilities are deployed is the warehouse. Maria Jesus Saenz, director of the MIT Digital Supply Chain Transformation Lab, and Benedict Jun Ma, postdoctoral associate at the MIT Digital Supply Chain Transformation
Lab, have researched this environment and the implications for the roles of humans and machines.
In many areas, such as picking items from racks and sorting them into bins, assistive technologies like pick-to-light systems have decreased the need for human expertise. Employees only have to follow computer instructions. However, there are value-adding tasks, such as processing order returns, that can require human judgment and problem-solving skills.
The degree of robot autonomy required in warehouses also varies. Low-autonomy machines, including automated guided vehicles, follow predefined paths. However, there are also high-autonomy robots at work in warehouses that utilize sophisticated sensors to plan their routes dynamically without human intervention. Given these variations and the diverse operating environments in facilities, the Digital Supply Chain Transformation Lab has developed a framework that helps managers optimize the mix of human and machine expertise in warehouses. It can also be used to help configure different combinations of skills and prepare managers to equip their facilities to meet future market demands.
The human-robot collaboration (HRC) framework is based on the degree of human expertise and robot autonomy involved in carrying out tasks. Situations where there are high levels of robot autonomy and limited human expertise (e.g., autonomous mobile robots), or vice versa (e.g., experienced humans performing value-added tasks), are referred to as Robot-in-the-Lead and Human-in-the-Lead configurations, respectively.
Examples of HRC in warehouses are at an early stage, according to the researchers, a notable example being collaborative order picking. An aspirational vision, known as Advanced HRC, is one where both human and machine elements are highly developed and integrated.
Regarding the use of AI, the researchers propose five key areas of impact for Advanced HRC.
For instance, Contextualization is where AI empowers robots to understand and adapt to their operational surroundings. An example is an AI-powered robot capable of automatically detecting changes in a warehouse layout and adjusting its movements accordingly. AI can also enhance the clarity and accuracy of robot responses by enabling seamless communication between machines and humans. An example is where human operators communicate with robots via voice commands. In another area, called Customization, AI helps to tailor robot behavior to humans’ skills and work routines.
Helping small retailers
Utilizing AI to enhance human performance is a central mission of the MIT Low-Income Firms Transformation (LIFT) Lab, led by Josué C. Velázquez from MIT CTL. However, in this case, the focus is on micro and small enterprises (MSEs) in developing countries. As well as playing a critically important economic role in local communities, MSEs represent a significant economic force worldwide. In Latin America and the Caribbean, for example, a region the LIFT Lab has prioritized, MSEs account for an estimated 99% of companies and 47% of employment. They also make up a sizable customer base for leading consumer products companies.
However, MSEs struggle to survive. Their mortality rate in developing countries is estimated to be over 30% annually during the first five years of operation. A key reason for this relatively short shelf life is their low productivity compared to larger firms. Store owners often lack education and training, and many use paper-based methods to manage their retail operations. Helping MSEs, particularly micro- and small-sized retailers, also known as nanostores, address
this productivity gap is one of the LIFT Lab’s primary goals.
The Lab has developed a groundbreaking chatbot called Lupita, an AI tool for nanostore owners and operators that is analogous to Amazon’s well-known virtual assistant, Alexa. Shop managers can interrogate Lupita and access a wealth of information such as the price of specific products in other retail outlets in the locale and details of their store’s inventory and delivery schedules. The chatbot supports critical store management functions like purchasing, report generation, inventory tracking, and sales monitoring.
Lupita provides shopkeepers with a powerful efficiency-building tool, including individuals who may have difficulty using off-the-shelf store management systems. Research conducted by the LIFT Lab in Mexico revealed that shopkeepers perceived Lupita as more efficient than established point-of-
sale systems.
The chatbot is a prime example of how AI can raise productivity by supplementing and enhancing human expertise.
AI as a change agent
Despite advances like these, developing workplace systems that facilitate AI-human working relationships is still in its early stages. Research at the MIT Digital Supply Chain Transformation Lab suggests that most warehouse operations today are in the “Elementary HRC” quadrant of the Lab’s framework (e.g., automated guided vehicles and humans performing tasks at stationary workstations), where humans and robots collaborate only on structured tasks. These collaborations must be taken to a higher level if facilities are to operate in the “Advanced HRC” quadrant.
Achieving such a transition should be an incremental process, the researchers suggest. Companies need to invest in systems that enable robots and humans to assume more sophisticated roles gradually. An example might be empowering robots with increased autonomy in repetitive tasks, such as sorting and picking, while training human workers to undertake more complex decision-making. Upskilling workers in this manner can also help prepare them for the future workplace. As Sheffi writes in The Magic Conveyor Belt, “AI and digital tools can augment the power of people, enabling them to handle jobs they could not in the past.”