In the more than two years since OpenAI introduced ChatGPT, generative AI (GenAI) has been mainly associated with changing how white-collar professions work. In a 2024 study by the Society for Human Resource Management and The Burning Glass Institute, the authors noted that the technology will broadly transform nearly all categories of white-collar work, while “blue-collar work will remain shielded from major disruptions.”

For now, maybe. But many industries typically associated with physical work are also experimenting with GenAI to resolve seemingly intractable issues. These range from improving crop yields in agriculture to easing labor shortages in construction.

What follows is a roundup of articles on how five such industries are putting GenAI to use. Their creative approaches offer lessons for any industry—white-collar or blue—on how to get the greatest value from this revolutionary technology.

Agriculture: Increasing farmers’ insights in a changing world

Sustainable food production is at the forefront of the agriculture industry’s challenges. Today’s farmers need to make efficient use of natural resources such as soil, seed, and water, as well as reduce the toll farming takes on the environment through methane generation, pesticide use, and other factors. At the same time, they must adapt to a changing climate that directly affects their revenue.

These pressures have prompted organizations to devise an array of GenAI applications, ranging from chatbots for small farmers in India, to data collection and analysis systems for multinational agricultural conglomerates.

Digital Green, a nonprofit that builds technology to help farmers in developing countries, has rolled out a virtual agronomist chatbot to extension agents in India, Kenya, and Ethiopia. The goal is to deliver information on farming essentials like seeds, climate, and even market prices.

And in Switzerland, tech company Datamars is investigating how GenAI could be used to improve milk production while also gathering sustainability data that is increasingly important, not only to the buying public, but also to governments

Cross-industry takeways

• Be ready to use a wide range of data for model training.
  Digital Green began its pilot by training the model with trusted scientific data from such sources as peer-reviewed agricultural research. But because farming is local, the next step will be to add regionally specific information like weather patterns and soil characteristics. Eventually, the organization hopes to use retrieval augmented generation to bring in continuously changing data, such as local market prices for various crops.
• Use experts to validate outputs.
  Initially, the people interacting with Digital Green’s chatbot are not farmers themselves but the agricultural extension agents who advise the farmers. The agents use the chatbot to get faster answers to farmers’ questions, and because of their agronomic expertise, they can spot and correct any GenAI hallucinations.
• Take measures to build user trust.
  In farming, it’s common for the same family to raise the same crops on the same land for generations. As a result, advice from outsiders, much less from a chatbot, may be unwelcome. To build trust, it’s essential to meet farmers on their own terms.
  
  To that end, the chatbot supports multiple languages and can respond to spoken questions rather than just relying on text. As the agricultural extension agents gradually familiarize farmers with the chatbot, the farmers will be able to use it in the field to ask questions such as, “What is the most profitable crop to grow, given seed and pesticide costs and local selling prices for certain crops or livestock?”

Construction: Combating labor and productivity challenges

Construction has historically been reluctant to invest in technology. But faced with labor and productivity challenges, major builders are experimenting with GenAI.

Skilled labor is growing scarce as older workers leave the industry and fewer young people enter its ranks. More than one in five construction workers are 55 or older, according to research by Associated Builders and Contractors. Meanwhile, productivity lags, and schedule and cost overruns are the norm, with just 8.5% of megaprojects ($1 billion or more) meeting or exceeding their time and budget expectations, according to Bent Flyvbjerg, emeritus professor at the University of Oxford’s Saïd Business School.

The hope with GenAI is that builders can use the accumulated decades’ worth of data to help solve these problems. “We’re sitting on 40 years of construction data,” says Kelsey Gauger, national director of operational excellence at Suffolk Construction. However, for many construction companies, that data is siloed, unstandardized, and hard to access.

Cross-industry takeaways

• Prioritize data organization and integration.
  Clark Construction is building an enterprise data warehouse and data fabric to connect its various IT systems. With this in place, it hopes to use GenAI to make better decisions and improve operations.
• Identify an urgent industry-wide problem.
  To help improve productivity—as well as train new hires more quickly—Suffolk Construction added a copilot to its collaboration software. The new feature makes it easier to find information, including answers to questions workers typically ask while on building sites.
• Look for quick yet substantial wins.
  Turner Construction, which manages 30,000 deals a year with its labor and materials suppliers, needed to boost the productivity of its procurement process. It is looking to use GenAI to process draft contracts by auto-generating text for statements of work or master service agreements. Gilbane Building Company, meanwhile, is piloting a large language model that helps project teams retrieve up-to-date construction documents within seconds.

Manufacturing: Easing skills shortages and avoiding costly downtime

Skilled workers are also in short supply in manufacturing. According to the Manufacturing Institute, the workforce development arm of the National Association of Manufacturers, a lack of skilled labor could leave 2.1 million manufacturing jobs unfilled by 2030.

This explains why one GenAI application that’s gaining ground in this industry is chatbots based on voluminous amounts of specifications, technical manuals, and other details. These bots can be used to train and provide answers in real time to junior-level staff, delivering instructions on how to fix equipment and even which tools are needed. This is crucial in an industry where downtime losses can total $532,000 an hour, on average.

Although interest in GenAI is high (some 78% of industrial manufacturing executives surveyed by KPMG named it as the top emerging technology), most are proceeding with caution. “Organizations are being cautious about applying GenAI to core operations that could directly impact their P&L,” says Kabali Ganesan, director of business consulting at Cognizant.

Cross-industry takeaways

• Choose low-risk, high-potential scenarios..
  An elevator manufacturer has developed a maintenance advisory system to improve aftermarket repairs. Based on user and service manuals for the company’s various elevator models and their major subsystems, a chatbot helps field technicians diagnose and resolve problems quickly. An early pilot reports a 90% reduction in repair times.
• Combine traditional AI with GenAI.
  According to Boston Consulting Group (BCG), manufacturers will need to employ both GenAI and machine learning/deep learning-based AI to improve their factory operations. For example, traditional AI can predict when a machine is in danger of failing, but when you add a GenAI-powered chatbot that can coach a maintenance engineer on taking action to avoid failure, it can be a force multiplier.
  
  “The chatbot could identify three things that may need adjustment or recalibration, then pull up the documentation showing the engineer how to do it,” says Randal Kenworthy, senior partner at West Monroe Partners. “Now you have a more informed and natural way of responding to the maintenance needs.”
• Don’t reinvent the wheel.
  Many manufacturers are relying on existing software vendors to fold GenAI capabilities into their existing platforms, according to the BCG report. Having GenAI already incorporated increases the likelihood that engineers will use the technology, Kenworthy says.

Aerospace and defense: Improving parts design

Manufacturers in aerospace and defense are especially conservative when it comes to implementing GenAI. Whether airplanes, satellites, drones, naval ships, or army tanks, the equipment they produce is extremely expensive, in the millions, even billions, of dollars. Just as important, the equipment must comply with strict military, government, and data security specifications.

While these companies see several potential benefits of GenAI, including improving predictive maintenance and aftermarket services, the most popular application today is in equipment design. Some 41% of aerospace and defense organizations surveyed by Capgemini said they were piloting generative AI in 3D modeling to speed up design, improve the aerodynamics of parts, and reduce costs.

Cross-industry takeaways

• Use GenAI to escape conventional (human) wisdom.
  Aerospace and defense companies have begun using what’s known as “generative design” to get a fresh perspective, sometimes going beyond what a human brain might have imagined. The concept made news in 2023 when a NASA engineer showed how the technology could design ultra-lightweight spacecraft parts in record time. At Boeing, researchers are exploring how GenAI models could improve aircraft design. “[AI] can put more complexity into its electronic brain than a human can; it can optimize over a broader space,” according to Todd Citron, Boeing’s chief technology officer.
• Look to the old, as well as the new.
  Using GenAI to write new code is a no-brainer, but experts suggest the technology could also help modernize legacy code. Aerospace and defense manufacturing operations are hampered by the legacy software upon which some embedded systems are still based. That makes system integration especially difficult.
  
  GenAI, though, could serve as a translator. “An interactive engine built on an underlying foundational model would be able to take the legacy code for a collision-avoidance module, for example, and revamp it into a modern programming language,” says Raman Ram, leader of EY America’s aerospace and defense practice.
• Take advantage of GenAI’s obsession with detail.
  If you’ve ever drafted a government contract, you may understand why the proverbial Department of Defense hammer costs $60,000. These contracts take significant time and expense to prepare because of the amount of detail and customization required. Companies can use GenAI to automate at least parts of the process, creating first drafts based on templates, historical documents, or specific prompts from procurement officials, according to a report by Deloitte’s AI Institute. Similarly, GenAI could help draft technical documentation, saving time and money.

Utilities: Improving infrastructure management and maintenance

Today’s utilities contend with myriad challenges, including dealing with many new sources of energy and evolving business models. Take West Penn Power, for example. Besides delivering a reliable source of electricity, the 100-year-old provider in Pennsylvania must now offer customers a percentage of electricity from alternative types of energy. It contracts with a couple dozen wind, solar, and hydropower suppliers, in addition to consumers who generate their own solar power and sell it back to West Penn.

GenAI could help companies sort out how to distribute which types of energy, and which supplier to pay for the energy used. The hope is that, by managing this complexity, GenAI will both reduce operating costs and increase reliability. Utilities are also using GenAI to improve maintenance and repairs, most notably by analyzing video of infrastructure. Nearly 40% of utility and energy companies surveyed by Capgemini have a dedicated team and budget for GenAI, and one in three have started pilots.

Cross-industry takeaways

• Look for areas of complexity that need simplification.
  The growing complexity in the utilities industry can be confusing for consumers, with the numerous energy options to choose from and various rates for each. With GenAI gathering the right data, customer service agents could answer questions and provide information about customer energy use, potentially in real time. They might even advise customers on how to save money by, for instance, identifying off-peak hours to charge their electric vehicles.
• Look far and wide, and down.
  Look beyond your four walls for GenAI benefits. Managing vegetation growth around power lines is a vital operation for reducing downtime and even avoiding wildfires. However, it’s slow and expensive—California’s Pacific Gas & Electric Co. spent approximately $2.5 billion over several years to identify and prune trees near power lines.
  
  Instead, utilities can use AI to identify the extent of vegetation near power line conductors, poles, or towers, and quickly target where intervention is needed. Many utilities already use camera-equipped drones to monitor outdoor physical infrastructure. However, the unstructured data they collect is impossible for humans to process in a reasonable amount of time, says Jing Wu, principal research director at the Info-Tech Research Group in Canada. With the right models and training, GenAI could quickly analyze this unstructured data and identify a defect or looming danger.
• Enhance your expertise.
  GenAI chatbots trained on the right manuals and data can help technicians in the field repair infrastructure more quickly and effectively. A maintenance worker, for instance, could ask a GenAI bot how to fix a specific piece of equipment on-site. After combing through data, the bot could provide instructions in plain language, easing access to expert knowledge.