CAREER FOCUS
Embracing AI as Engineering Evolves
Although artificial intelligence cannot replace engineers, the technology’s role across industries will force changes to which engineers must adapt.
Written by Robin Flanigan
ARTIFICIAL INTELLIGENCE CAN EXPEDITE research and development, improve documentation, elevate design processes, and more—all tasks relevant to a mechanical engineer’s work.
So, should mechanical engineers be worried about being replaced by ever-emerging AI technology?
“They should not be worried, but they should be aware that they might need to adapt,” said Brenna Argall, professor of mechanical engineering at Northwestern University. “AI won’t take over their job, but it might change the way they do their job.”
Mechanical engineer employment is projected to grow 11 percent from 2023 to 2033—much faster than the average for all occupations, according to the U.S. Bureau of Labor Statistics.
Knowing how to work with AI will be an increasingly desirable qualification in the job market moving forward. As many 71 percent of corporate leaders say they would prefer a less-experienced candidate with AI skills over a more experienced one without them, according to the 2024 Work Trend Index by Microsoft and LinkedIn.
Since generative AI is seen most prevalently in text, code, and image generation, “it can be a tool you leverage to do your work more efficiently,” Argall said.
For example, modeling tasks can be powered more proficiently using AI, which looks different from automation in multiple ways. The robotics research community has been framing this difference around how perceptions of the world impact decision making.
“Instead of having a closed-loop feedback control—that often happens in automation—you are changing your decision-making to a higher, more sophisticated level,” Argall explained. “You now are interpreting sensor data and have a perception system within your autonomy system.”
Even so, AI-powered chatbots cannot be relied on to tackle complex coding problems. At this point, it can only write “pretty good code,” and that’s if you know what kind of prompts to feed it—and as long as what it is being asked to generate is not terribly elaborate, Argall explained.
“There are still things that humans are better at. And from a more philosophical standpoint, language, code, and image generation are not the only markers of intelligence.”
—Brenna Argall, professor of mechanical engineering, Northwestern University
“It’s very important to understand what these recent advances can do, and what they can’t do—and why,” Argall continued. “That ‘why’ really informs whether it’s remotely on the horizon that they will translate over to spaces that mechanical engineers are concerned with.”
Automotive manufacturing, where large, powerful, very precise robotic arms have been operating the same way for decades, is a classic example. As Argall described it, those robotic arms expect the world to be fixed—a specific part will be in a specific place—and are made to shut down when faced with uncertainty or deviation. A more sophisticated perception system, however, may be able to recognize a particular part regardless of where it is located. So, if a human hands the robotic arm the wrong part by mistake, an algorithm using AI can know that and adjust as needed.
“You can now have a much richer interaction and a lot more flexibility,” Argall added. “You will probably be less fast, but you also won’t need to have a single-purpose setup where one robotic arm is only doing one thing.”
This type of progress is opening manufacturing opportunities for small- and medium-sized enterprises historically unable to afford robotics because of this single-purpose setup.
Meanwhile, as AI becomes more prevalent, it’s important to consider what goes into machine learning algorithms used to analyze data and make predictions. While numerous toolboxes exist to take care of the writing and deployment of data for AI algorithms, Argall urged contemplating these questions: What examples are you giving it? What examples are you leaving out? Where might you be introducing bias or noise in your data?
“You also need to understand the algorithms well enough to know where they might introduce bias as well—and is that okay?” she said. “As this data curation side of machine learning gets used for artificial intelligence, this is critical for domain experts.”
While AI can automate routine tasks, allowing for more creative and strategic work, and while humans and robots have more chances to collaborate, make no mistake: “There are still things that humans are better at,” Argall said. “And from a more philosophical standpoint, language, code, and image generation are not the only markers of intelligence.”
To make peace with how AI can help mechanical engineers stay competitive and relevant, Argall recommended thinking about how computers came onto the work scene in the 1960s and 1970s and now are ubiquitous.
“It’s not that we have fewer engineers as a result, but the way engineers do their work did change,” she said. “Anticipate that a similar change might be on the horizon—and it will be okay.”
Robin Flanigan is an independent writer in Rochester, N.Y.
© 2025 The American Society of Mechanical Engineers. All rights reserved.