Artificial intelligence has rapidly transformed software engineering, with generative AI and large language models (LLMs) now capable of producing vast quantities of code and documentation, while machine-learning algorithms excel at monitoring performance and detecting security vulnerabilities. However, the application of these powerful AI tools to the complex, safety-critical domain of physical system design, such as a jet engine, has remained a less explored frontier. The MIT JARVIS Challenge (Jet-engine AI Research and Validation Intensive Sprint) recently sought to bridge this gap, tasking undergraduate students with exploring AI’s potential to compress the design-build-test cycle for a small gas turbine engine, ultimately revealing both the profound capabilities and critical limitations of AI in hardware engineering.
The Genesis of JARVIS: A Quest for Accelerated Innovation
The JARVIS Challenge emerged from a growing recognition within the engineering community that while AI has revolutionized digital domains, its impact on the physical world, particularly in fields demanding rigorous safety and precision like aerospace, was still largely theoretical. Traditional aerospace development cycles are notoriously long, often spanning decades and involving billions of dollars, even for incremental design iterations. The challenge aimed to determine if AI could dramatically reduce these timelines, enabling faster innovation and iteration in complex hardware.
Professor Zolti Spakovszky, director of the MIT Gas Turbine Laboratory, articulated the challenge’s core finding: "The JARVIS challenge showed that AI can substantially accelerate safety-critical hardware engineering, but engineering judgment remains the decisive differentiator. An AI-native engineer is not defined by using AI, but by leading it — knowing when to trust it, when to challenge it, and how to translate AI outputs into working hardware. Manufacturing — not engineering design or analysis — remained the fundamental rate-limiting step." This statement underscores a critical insight: AI is a powerful tool, but the human engineer remains the ultimate arbiter of success, especially in domains where failure carries severe consequences.
Challenge Parameters: Building a Jet Engine in Weeks
The JARVIS Challenge was an intensive four-week sprint designed to push the boundaries of AI-assisted engineering. Undergraduates were given the formidable task of designing, fabricating, assembling, and testing a small gas turbine aero engine. The specific objective was to construct a "JARVIS-class" single-spool jet engine capable of producing 50–100 pounds of thrust, operating on Jet-A fuel, and completing five 60-second runs. Teams were granted complete autonomy over their design choices, materials, and fabrication methods, fostering an environment of true innovation and problem-solving.
Thirty-one students, representing nearly every department within the MIT School of Engineering, formed seven distinct teams. Their experience levels varied significantly, from all-first-year groups to those dominated by seniors. Many participants initially possessed limited knowledge of turbomachinery, compressible flows, or even fundamental thermodynamics. For some of the younger students, this was their first encounter with the internal workings of a gas turbine. This diverse skill set and experience level provided a unique testing ground for how AI could democratize complex engineering tasks.
To support their ambitious endeavor, the teams had access to a comprehensive suite of resources. These included MIT’s advanced machine shops and a network of manufacturing vendors, commercial engineering software such as Concepts NREC for turbomachinery design, SolidWorks for CAD, and ABAQUS for finite element analysis. Crucially, they also had access to MIT Parley, a newly launched platform that aggregates frontier large language models through a single, unified interface. This allowed JARVIS leads to monitor student interactions with AI tools, including prompt usage, cost per prompt, and specific LLMs employed. With early access to Parley and substantial financial backing from MIT Lincoln Laboratory, the Department of Mechanical Engineering, and corporate sponsors like Safran, Voyager Technologies, Beehive Industries, and Boom Technology, students benefited from essentially unlimited AI usage.
The involvement of corporate sponsors was driven by both recruiting interest and a genuine curiosity about how AI might reshape future engineering workflows. Ryan (Hal) Hefron of Voyager Technologies emphasized the long-term vision, stating to 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 the competition with keen interest: "JARVIS was a genuine experiment, a learning endeavor. We frankly didn’t know what to expect, from the students or from the AI models. What struck me coming from the students was: first, the enthusiasm to explore; then, as the project developed, they all came to the cool-headed realization of what AI could or could not help them with, and then almost instantly adapted for that." Garnier’s observation points to the rapid adaptability and critical thinking demonstrated by the students, suggesting a promising future for AI-native engineers.
The faculty leadership, comprising professors Zachary Cordero, Zolti Spakovszky, Masha Folk, and Andreea Bobu from the Department of Aeronautics and Astronautics, alongside engineers from Lincoln Laboratory and a dedicated team of teaching assistants, provided oversight, particularly regarding safety. Weekly progress reviews allowed them to critically evaluate student progress and assess their AI utilization strategies. Professor Spakovszky employed a subtle guiding technique, often posing questions like, "Do you know what a rabbet fit is? Take in the comment," to prompt critical thinking without directly providing solutions.
A Four-Week Sprint: AI as Co-Pilot, or Obstacle?
The JARVIS Challenge unfolded as a dynamic four-week sprint, offering a real-time crucible for AI’s utility in accelerated hardware design.
Week 1: Conceptualization and Knowledge Acquisition.
The initial week saw teams grappling with the foundational concepts of jet engine design. One team withdrew early, highlighting the intense pressure and complexity of the task. The remaining teams, with varying degrees of success, developed preliminary designs for their gas turbines. AI tools proved immediately valuable in this phase, serving as powerful information retrieval and synthesis agents. Students leveraged AI to summarize dense textbooks on turbomachinery, learn to operate complex design software, identify potential manufacturing vendors, generate structured Excel sheets for data management, answer specific technical questions, find academic references, and conduct comparative analyses between different design decisions. One particularly innovative team even created an AI agent within Parley, entrusting it with project management responsibilities. This initial success demonstrated AI’s capacity to significantly accelerate the knowledge acquisition and conceptual design phases, particularly for students new to the field.
Week 2: Detailed Design and Prototyping.
As the challenge progressed into its second week, teams transitioned to more intricate tasks, including detailed CAD design, ordering components, and prototyping combustors. It was during this phase that the limitations of current AI capabilities began to surface. While LLMs like Claude and ChatGPT were adept at suggesting design alternatives and filling theoretical knowledge gaps, students increasingly encountered "hallucinations" — factually incorrect or nonsensical outputs presented as truth — along with "sycophancy," where AI would agree with or reinforce user assumptions without critical evaluation. This lack of inherent physical understanding and reliable validation from the AI tools started to erode student confidence and, paradoxically, slowed down their progress.
Elizabeth Tupaj, a member of the 811 Crew team, articulated this sentiment: "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." This firsthand experience highlighted the critical need for human oversight and expertise to filter and validate AI-generated content, especially in safety-critical applications. Teaching assistant John Zhang observed a similar pattern: "Seeing this firsthand with the students reminded me how much first impressions matter. 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." This underscores the importance of reliable AI performance, particularly in early interactions, to build user trust and foster continued adoption.
Final Weeks: Fabrication, Assembly, and Testing.
In the culminating weeks, the remaining finalist teams encountered an obstacle that no AI could solve: the complexities of working with human vendors and the realities of supply chain management. Students reported that while AI searches could identify numerous potential vendors, these entities often lacked existing rapport with the student teams and had little interest in accommodating their extremely tight timelines. The crucial lesson learned was that "the vendors who came through were the ones our team had personal relationships with." This experience profoundly illustrated that human networks, trust, and established relationships remain indispensable in the practical execution of engineering projects, particularly under pressure.
Despite these hurdles, the teams pressed forward. Of the three finalists, only "Fast and Fractured" achieved first-attempt ignition of their mini-combustor. This team had extensively used AI for trade studies and architectural comparisons, successfully arriving at a viable design despite none of its members having prior gas turbine experience. Professor Masha Folk, the Charles Stark Draper Career Development Professor of Aeronautics and Astronautics, remarked on this pivotal moment: "The JARVIS Challenge showed what’s possible when you combine AI-enabled design with motivated students and a culture of rapid experimentation. The moment that stood out most was when 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 while giving students hands-on experience with a real engineering challenge." This successful ignition served as a powerful validation of AI’s potential to accelerate initial design and prototyping, even if other stages presented their own challenges.
The Verdict: Human Judgment Prevails
By the end of May, the two most senior teams, "Fast and Fractured" and "811 Crew," had completed full engine tests. "Fast and Fractured," despite their AI-assisted design, faced repeated delays due to vendor issues but eventually reached the test stand. Unfortunately, their hot fire test was cut short when the rotor rubbed and seized against the stationary housing, a common failure mode in turbomachinery. Team "811 Crew," however, emerged victorious. With more prior exposure to turbomachinery and propulsion concepts, their engine started successfully, transitioned to Jet-A fuel, and generated net thrust.
PhD student Joe Chiapperi vividly recalled the intensity of the final tests: "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! What these students accomplished in such a short time span is nothing short of amazing."
Interestingly, the "811 Crew" had been notably resistant to heavy AI usage throughout the competition, relying instead on their fundamental engineering knowledge and robust teamwork. Elizabeth Tupaj elaborated on their strategy: "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." This highlights a significant finding: while AI offers substantial support, a strong foundation in core engineering principles and domain-specific knowledge can be a more direct path to success, especially when AI tools are still maturing.
A clear pattern emerged from the JARVIS Challenge regarding AI adoption: younger students, often with less prior engineering experience, tended to use Parley more frequently and creatively, leveraging AI to fill knowledge gaps. In contrast, juniors and seniors, possessing deeper foundational expertise, relied more on their established understanding. Professor Andreea Bobu synthesized this observation: "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. To me, that’s the real opportunity ahead: training the next generation of engineers who have the judgment to direct these AI tools and the instinct to reach for them."
The competition’s most salient finding was that engineering experience acts as a multiplier. The human factor, particularly sound engineering judgment derived from mastering first principles and fundamental concepts, proved vital in navigating complex decisions amidst incomplete information. When it comes to designing and building safety-critical physical systems, human hands, critical thinking, and ultimate accountability remain irreplaceable. Teaching assistant Kyle Woody summarized this succinctly: "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."
Beyond the Workshop: Implications for AI-Native Engineering
The implications of the JARVIS Challenge for the future of aerospace and broader engineering are significant. If small, agile teams, effectively utilizing AI copilots, can compress design-build-test cycles from years to mere weeks, the consequences for workforce structure, research and development timelines, and global competitive dynamics could be revolutionary. For context, the global market for AI in engineering is projected to reach tens of billions of dollars within the next decade, driven precisely by the efficiencies demonstrated in challenges like JARVIS. Traditional jet engine development, for instance, typically involves hundreds of engineers over many years, with component design alone often taking months or even a year for critical parts. The JARVIS challenge compressed this into a four-week sprint for a complete engine system, albeit a smaller one.
The students who participated in the JARVIS Challenge are at the vanguard of a new era of "AI-native engineering." They grappled with these stakes not as a theoretical exercise but as a tangible reality, with tangible hardware on a test stand. Their experiences will inform how future generations of engineers integrate AI into their workflows, moving beyond simple tool usage to a symbiotic partnership.
Associate director of the MIT Gas Turbine Laboratory, Professor Zachary Cordero, reflected on the broader educational takeaway: "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 like MIT Motorsports and Rocket Team. Performance in JARVIS correlated strongly with year in school. My main takeaway is that in the AI era, education is more valuable than ever." This emphasizes that AI, rather than diminishing the need for fundamental knowledge, elevates it. A deeper understanding of first principles allows engineers to critically evaluate AI outputs, identify errors, and guide the tools effectively, transforming AI from a potential source of misinformation into a powerful accelerator of human ingenuity. The JARVIS Challenge has, therefore, not only illuminated the path forward for AI in hardware engineering but also reinforced the enduring and irreplaceable value of human expertise and critical judgment.
