Artificial intelligence has undeniably revolutionized software engineering, with generative AI and large language models (LLMs) capable of producing vast quantities of code and documentation, while machine learning algorithms excel at performance monitoring and security vulnerability detection. However, the true transformative potential of these AI tools when applied to the complex, physical realm of conceiving, designing, and building intricate systems such as jet engines has remained a subject of intense exploration. This past semester, the JARVIS Challenge, an intensive sprint focused on Jet-engine AI Research and Validation, sought to answer this critical question, tasking MIT undergraduates with discovering whether AI could significantly accelerate and enhance the design-build-test cycle for hardware engineering.
The outcome of the JARVIS Challenge, according to Professor Zolti Spakovszky, director of the MIT Gas Turbine Laboratory, was a compelling demonstration that AI can indeed substantially accelerate safety-critical hardware engineering. However, he emphasized that "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."
The Challenge: A Four-Week Sprint to Jet Engine Innovation
The core objective of the JARVIS Challenge was ambitious: to compress the traditional, often multi-year, design-build-test cycle of a complex piece of machinery into a mere four weeks. Undergraduates were tasked with designing, fabricating, assembling, and testing a small gas turbine aero engine, with AI serving as their primary engineering partner. The target was to construct a "JARVIS-class" single-spool jet engine capable of producing between 50 and 100 pounds of thrust, fueled by Jet-A, and successfully completing five 60-second runs. Teams were granted complete autonomy over their design choices, material selection, and fabrication methods.
This rigorous challenge attracted a diverse cohort of 31 students, representing nearly every department within MIT’s School of Engineering. They organized themselves into seven distinct teams, with compositions ranging from all first-year students to groups heavily populated by seniors. A significant number of participants entered the competition with limited prior experience in specialized fields such as turbomachinery, compressible flows, or, for the younger students, even fundamental thermodynamics. Many had never physically encountered a gas turbine before committing to building one from scratch.
Tools and Resources at Their Disposal
To support this ambitious endeavor, the student teams were provided with access to MIT’s state-of-the-art machine shops and a network of manufacturing vendors. They also had access to industry-standard commercial software packages, including Concepts NREC for turbomachinery design, SolidWorks for 3D modeling, and ABAQUS for finite element analysis. Additionally, various specialized test rigs were available for characterizing and assembling individual engine components.
A key enabler of the challenge was the early access granted to MIT Parley, a newly launched platform designed to aggregate frontier large language models through a unified interface. This provided the JARVIS leadership with unprecedented visibility into the students’ AI usage, allowing them to track prompts, associated costs, the specific LLMs employed, and other crucial operational data. With financial backing from MIT Lincoln Laboratory, the Department of Mechanical Engineering, and corporate sponsors Safran, Voyager Technologies, Beehive Industries, and Boom Technology, the students enjoyed virtually unlimited access to AI resources.
The involvement of these sponsors was driven by a dual interest: a keen desire for recruitment opportunities and a genuine curiosity about the disruptive potential of AI in reshaping engineering workflows. Ryan (Hal) Hefron of Voyager Technologies articulated this sentiment, telling 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, noting, "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." He added, "It makes me confident that this generation of leading engineers will probably not fall prey to easy and shortsighted use of AI, and will do so by keeping ever more in contact with experiments – physical or thought experiments."
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, ensured the safety and integrity of the project. Through weekly progress reviews, they critically evaluated student advancements and their integration of AI tools. Professor Spakovszky developed a nuanced approach to guiding teams, posing probing questions rather than providing direct solutions, such as inquiring, "Do you know what a rabbet fit is? Take in the comment."
AI’s Dual Role: Facilitator and Foe
By the conclusion of the first week, one team had withdrawn from the competition, while the remaining groups had, with varying degrees of success, formulated initial designs for their gas turbines. AI was employed by different teams for a range of tasks, including summarizing technical literature, learning to operate design software, identifying potential vendors, generating comparative analysis for design decisions, answering specific technical queries, and sourcing relevant research references. One particularly innovative team even created an AI agent within Parley to function as their project manager.
As teams moved into the second week, the focus shifted to developing detailed CAD designs, ordering components, and prototyping critical elements like combustors. It was at this stage that the limitations of AI began to surface more acutely. While LLMs like Claude and ChatGPT proved adept at suggesting design alternatives and bridging knowledge gaps, the notorious tendencies of generative AI – including hallucinations, sycophancy, and a lack of inherent physical understanding – began to undermine student confidence and decelerate progress.
Elizabeth Tupaj, a member of the "811 Crew" team, reflected on this experience: "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 observed, "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."
The final weeks of the challenge presented another significant hurdle that no AI could surmount: navigating the complexities of vendor relationships. Students reported that "AI searches found vendors we had no rapport with, who had no interest in our tight timeline. The vendors who came through were the ones our team had personal relationships with." This underscored the enduring importance of human connection and established networks in the practical execution of engineering projects.
Of the three teams that reached the finals, only "Fast and Fractured" achieved first-attempt ignition of their mini-combustor. This team had leveraged AI extensively for trade studies and architectural comparisons, ultimately arriving at a viable design despite none of its members possessing prior gas turbine experience.
Masha Folk, the Charles Stark Draper Career Development Professor of Aeronautics and Astronautics, highlighted the success of this approach: "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."
The Vanguard of AI-Native Engineering
By the end of May, two of the more senior teams – "Fast and Fractured" and "811 Crew" – had completed full engine tests. "Fast and Fractured," despite their AI-assisted design prowess, faced persistent delays due to vendor issues but eventually reached the testing phase. Their hot fire run was unfortunately cut short when the rotor experienced a rub and seized against the stationary housing. In contrast, the "811 Crew," who had greater prior exposure to turbomachinery and propulsion concepts, emerged as the victorious team. Their engine successfully started, transitioned to Jet-A fuel, and generated net thrust.
PhD student Joe Chiapperi described the palpable tension 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 winning "811 Crew" team had initially been resistant to heavily relying on AI throughout the competition, preferring to trust their foundational knowledge and teamwork. Tupaj explained, "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."
A notable observation from the challenge was the differing approaches to AI utilization based on student experience. Younger students, less encumbered by preconceived notions, tended to use Parley more frequently and innovatively. Conversely, juniors and seniors often leveraged their deeper domain expertise.
Professor Andreea Bobu offered a critical insight into the findings: "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."
Key Takeaways and Future Implications
The competition’s most definitive finding was that engineering experience acts as a significant multiplier, and the human element remains indispensable. A strong grasp of first principles and fundamental concepts fosters sound engineering judgment, enabling individuals to navigate complex decisions amidst incomplete information. Crucially, when it comes to constructing safety-critical physical systems, human oversight and accountability are irreplaceable.
Teaching assistant Kyle Woody summarized this point: "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."
The broader implications of AI in aerospace are profound. If small teams, empowered by well-managed AI copilots, can condense design-build-test cycles from years to mere weeks, the ramifications for workforce structure, research and development timelines, and competitive dynamics within the industry could be substantial. The students who participated in the JARVIS Challenge are at the forefront of this paradigm shift, grappling with these high stakes not as a theoretical exercise, but within the tangible environment of a machine shop, with a jet engine humming on a test stand.
Professor Cordero, associate director of the MIT Gas Turbine Laboratory, concluded, "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 sentiment underscores the ongoing need to cultivate engineers who can not only utilize AI tools effectively but also critically evaluate their outputs and lead the integration of AI into the future of complex system design.
