Artificial intelligence has rapidly reshaped software engineering, but can it compress the design-build-test cycle of a physical system as complex as a jet engine? The JARVIS Challenge (Jet-engine AI Research and Validation Intensive Sprint) at MIT put that question to a rigorous test this past semester, and the results offer a nuanced answer: AI can dramatically accelerate safety-critical hardware engineering, but human judgment remains the decisive differentiator. The competition, which gave undergraduates four weeks to design, fabricate, assemble, and test a small gas turbine aero engine, revealed that AI-native engineering is not about using AI—it is about leading it, knowing when to trust its outputs, when to challenge them, and how to translate those outputs into working hardware.
What the JARVIS Challenge Set Out to Prove
The JARVIS Challenge asked 31 MIT undergraduates organized into seven teams to build a single-spool jet engine producing 50–100 pounds of thrust, running on Jet-A fuel, and completing five 60-second runs. Teams had total freedom over design, materials, and fabrication. Representing nearly every department in the School of Engineering, the students ranged from first-years with little to no experience in turbomachinery or thermodynamics to senior-heavy groups with deeper technical backgrounds. Many had never seen the inside of a gas turbine before signing up to build one.
At their disposal: MIT’s machine shops and manufacturing vendors; commercial software including Concepts NREC, SolidWorks, and ABAQUS; and various test rigs for characterizing and assembling components. Crucially, teams had access to MIT Parley, a newly launched platform that aggregates frontier large language models through a single interface. Through Parley, JARVIS leads could see how students were using the AI tools—their prompts, the cost per prompt, the specific LLMs being used, and other critical information. With financial support from MIT Lincoln Laboratory, the Department of Mechanical Engineering, and corporate sponsors Safran, Voyager Technologies, Beehive Industries, and Boom Technology, students had essentially unlimited access to AI.
The sponsors were drawn by recruiting interest and genuine curiosity about how AI might reshape engineering workflows. Ryan (Hal) Hefron of Voyager Technologies told the students, “We see this as the future of engineering. You’re honing skills that are not just nice to have—they’re going to be the future baseline in the engineering workforce.” Vincent Garnier, managing director of Safran Tech, observed that the students moved from enthusiasm to a cool-headed realization of what AI could and could not help them with, and then adapted almost instantly. “It makes me confident that this generation of leading engineers will probably not fall prey to easy and shortsighted use of AI,” he said.
How Teams Used AI—and Where It Fell Short
By the end of week one, one team had withdrawn. The others had, with varying degrees of success, developed initial designs. Teams used AI to summarize textbooks, teach themselves design software, source vendors, create Excel sheets, answer specific questions, find references, and perform comparative analysis between design decisions. One team created an agent in Parley and tasked it with serving as their project manager.
By week two, teams began detailed CAD design, ordering parts, and prototyping combustors. This is where the limitations of AI became apparent. While Claude and ChatGPT were adept at offering design alternatives and filling knowledge gaps, teams found that hallucinations, sycophancy, and the lack of physical understanding inherent in generative AI undermined their confidence and slowed them down. Elizabeth Tupaj of team 811 Crew put it directly: “AI is a helpful tool, great at finding information, helping organize things, and can write well, but it can’t do design. The moment the engineer doesn’t know what is going on and the AI is in charge is the moment the design becomes unreliable, at least with AI at its present capabilities.”
Teaching assistant John Zhang noted that first impressions mattered significantly. “If the students couldn’t get answers from the AI early on, they quickly grew frustrated and formed a lasting opinion that precluded them from using it later.”
In the final weeks, the finalists hit another obstacle no AI could solve: working with vendors. “AI searches found vendors we had no rapport with, who had no interest in our tight timeline,” students reported. “The vendors who came through were the ones our team had personal relationships with.”
Of the three finalists, only Fast and Fractured achieved first-attempt ignition of their mini-combustor. The team had used AI heavily for trade studies and architecture comparisons, arriving at a viable design despite none of them having prior gas turbine experience. Masha Folk, the Charles Stark Draper Career Development Professor of Aeronautics and Astronautics, described the moment: “The first student-designed combustor was installed on the test stand. It ignited flawlessly, ramped to full power, transitioned to dual-fuel operation, and then sustained stable combustion on 100 percent Jet-A fuel. This was proof that we can dramatically accelerate the cycle of design, build, and test.”
The Winning Team Trusted Fundamentals Over AI
By the end of May, the two more senior teams—Fast and Fractured and 811 Crew—had completed full engine tests. Fast and Fractured, with their AI-assisted design, were delayed by vendor issues week after week. Their hot fire was cut short when the rotor rubbed and seized against the stationary housing. Team 811 Crew, who had more exposure to turbomachinery and propulsion concepts going into the competition, emerged victorious. Their engine started, successfully transitioned to Jet-A, and generated net thrust.
PhD student Joe Chiapperi described the moment: “As we stood there with the air-starter, hearing their engines spool up and watching them spit fire, it felt like my heart was racing out of my chest. There were so many ways it could go wrong!”
The 811 team had been resistant to using AI throughout the competition, trusting instead to their fundamentals and teamwork. “We had people who were at least somewhat familiar with the design software, mechanical engineers who knew how to build anything, and aerospace engineers who had taken classes on the design of gas turbine engines specifically,” said Tupaj.
From the start, younger students used Parley more frequently and creatively, while the juniors and seniors leveraged deeper experience. This pattern revealed a critical insight. Professor Andreea Bobu summarized it: “JARVIS taught me that getting value from AI takes two things: enough expertise to judge what it tells you and catch it when it’s wrong, and enough curiosity to actually lean on it where it could help. The team that moved fastest in the sprint was experienced and leaned heavily on AI to get there. The team that eventually won was more resistant to AI; they had the expertise, but that skepticism made them slower. The sweet spot seems to be knowing enough to stay in charge of the tool, and being eager enough to pick it up in the first place.”
What Is the JARVIS Challenge and What Did It Reveal About AI in Engineering?
The JARVIS Challenge was a four-week intensive sprint at MIT that asked undergraduate teams to design, build, and test a working jet engine using AI as a primary engineering partner. The key finding: AI can compress design cycles and accelerate analysis, but engineering judgment—rooted in first-principles knowledge and hands-on experience—remains essential. Teams that combined deep expertise with strategic AI use moved fastest, while the winning team relied on fundamentals and teamwork over AI. The challenge demonstrated that manufacturing, not engineering design or analysis, is the fundamental rate-limiting step in hardware development, and that no AI can replace the human relationships and accountability required to build safety-critical physical systems.
Engineering Experience as a Multiplier in the AI Era
The competition’s clearest finding is that engineering experience acts as a multiplier for AI productivity. Mastering first principles and fundamental concepts breeds good engineering judgment and the ability to navigate tough decisions in the face of incomplete information. Teaching assistant Kyle Woody noted, “JARVIS has shown that AI copilots can have a multiplicative effect on engineering productivity, with judgment and first-principles thinking serving as the key differentiators among teams.”
Professor Zolti Spakovszky, director of the MIT Gas Turbine Laboratory, put it even more starkly: “Manufacturing—not engineering design or analysis—remained the fundamental rate-limiting step.” This is a crucial distinction for any organization considering AI in hardware development. The AI tools available today excel at generating design alternatives and filling knowledge gaps, but they cannot negotiate with suppliers, operate a lathe, or take responsibility for a safety-critical system.
The Implications for Aerospace and Engineering Workforce
The implications of JARVIS for aerospace engineering are significant. If small teams using well-managed AI copilots can compress design-build-test cycles from years to weeks, the consequences for workforce structure, R&D timelines, and competitive dynamics could be substantial. The students who tackled the JARVIS Challenge are among the first engineers to grapple with those stakes not as a thought experiment, but in a machine shop, with a jet engine on the test stand.
Professor Zachary Cordero, associate director of the MIT Gas Turbine Laboratory, emphasized the educational angle: “JARVIS highlighted the power of AI in the design of physical systems. But it also showed that the key to unlocking that power is education, through coursework, internships, and hands-on extracurriculars. Performance in JARVIS correlated strongly with year in school. My main takeaway is that in the AI era, education is more valuable than ever.”
What This Means for Engineers and Organizations
The JARVIS Challenge offers a concrete, evidence-based picture of AI’s role in hardware engineering today. AI copilots can accelerate the early stages of design, help teams explore trade-offs, and fill knowledge gaps—especially for less experienced engineers. But the technology remains unreliable for detailed design work, lacks physical understanding, and cannot replace the human judgment that comes from hands-on experience and first-principles thinking. The teams that succeeded were those that led their AI tools, not those that followed them.
For organizations looking to integrate AI into engineering workflows, the lesson is clear: invest in education and hands-on experience alongside AI deployment. The engineer who knows enough to direct the tool and catch its mistakes will always outperform the engineer who relies on the tool to lead. The next generation of AI-native engineers will be defined not by the models they use, but by the judgment they bring to the work—and no amount of compute can substitute for that.
