Artificial intelligence (AI) has rapidly redefined the landscape of software engineering, with generative AI and large language models (LLMs) demonstrating unprecedented capabilities in code generation, debugging, and documentation. Machine learning algorithms are increasingly deployed to monitor software performance and detect security vulnerabilities with remarkable efficiency. However, a critical question has persisted regarding the extent of AI’s transformative potential when applied to the complex, iterative, and physically constrained process of conceiving, designing, fabricating, and testing intricate physical systems, such as a modern jet engine. This query formed the bedrock of the JARVIS Challenge (Jet-engine AI Research and Validation Intensive Sprint), an intensive academic endeavor undertaken at the Massachusetts Institute of Technology (MIT) this past semester, aiming to ascertain whether AI could significantly compress the traditionally protracted design-build-test cycle in hardware engineering.
A Bold Experiment in AI-Native Engineering
The JARVIS Challenge was a crucible for a new generation of engineers, tasking MIT undergraduates with exploring the efficacy of AI as a primary engineering partner in an ambitious, time-bound project. The objective was clear: design, fabricate, assemble, and test a small gas turbine aero engine within an aggressive four-week timeframe. Professor Zolti Spakovszky, director of the MIT Gas Turbine Laboratory and a leading voice in turbomachinery, encapsulated the challenge’s profound findings: "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: while AI offers immense leverage, the human element of discernment, experience, and practical application remains indispensable, particularly in domains where physical reality imposes unforgiving constraints.
The design-build-test cycle for a new aircraft engine, even a component, can traditionally span years, if not decades, involving hundreds of engineers, vast computational resources, and extensive physical prototyping. The very notion of compressing this into a single academic semester, let alone four weeks, was audacious. It represented a direct confrontation with the established paradigms of aerospace engineering, pushing the boundaries of what is conventionally considered achievable within such a constrained timeline.
The Challenge Blueprint: Teams, Tools, and Task
The JARVIS Challenge was meticulously structured to simulate a real-world, high-pressure engineering sprint. Thirty-one MIT undergraduates, representing a diverse array of departments across the School of Engineering, formed seven distinct teams. These teams ranged from groups composed entirely of first-year students to those predominantly populated by seniors, reflecting a spectrum of academic experience. Remarkably, many participants initially possessed limited to no background in specialized fields critical to gas turbine design, such as turbomachinery, compressible fluid dynamics, or even fundamental thermodynamics. Many had never encountered the internal workings of a gas turbine before embarking on the task of building one. This deliberate inclusion of students with varying levels of prior knowledge was crucial for observing how AI tools might bridge knowledge gaps and democratize access to complex engineering domains.
Their mandate was to construct a "JARVIS-class" single-spool jet engine capable of producing 50-100 pounds of thrust, operating on Jet-A aviation fuel, and successfully completing five 60-second runs. The teams were granted complete autonomy over design choices, material selection, and fabrication methods, fostering an environment of true innovation and problem-solving.
To support their ambitious undertaking, students were provided with an impressive arsenal of resources. Access to MIT’s state-of-the-art machine shops and a network of manufacturing vendors was crucial for translating digital designs into physical components. Commercial software suites, including Concepts NREC for turbomachinery design, SolidWorks for detailed CAD modeling, and ABAQUS for finite element analysis, provided industry-standard tools for advanced engineering. Furthermore, various test rigs were available for characterizing individual components and facilitating assembly.
A pivotal technological enabler was MIT Parley, a recently launched institutional platform that aggregates frontier large language models through a single, unified interface. Parley granted JARVIS Challenge organizers unprecedented visibility into student workflows, allowing them to monitor prompts, track associated costs, identify specific LLMs utilized, and glean other critical data. Early access to Parley, coupled with substantial financial backing from MIT Lincoln Laboratory, the Department of Mechanical Engineering, and corporate sponsors — including Safran, Voyager Technologies, Beehive Industries, and Boom Technology — ensured students had virtually unlimited access to AI resources. These corporate sponsors were motivated not only by recruiting interests but also by a genuine, forward-looking curiosity about how AI might fundamentally reshape future engineering workflows and accelerate product development. Ryan (Hal) Hefron of Voyager Technologies articulated this sentiment directly 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, a global leader in aerospace propulsion, closely observed the challenge’s progression. He noted, "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. 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." This observation highlights the development of critical thinking and adaptive strategies among the student engineers, a testament to their emergent "AI-native" capabilities.
Faculty leadership, including Professors Zachary Cordero, Zolti Spakovszky, Masha Folk, and Andreea Bobu from the Department of Aeronautics and Astronautics, along with engineers from MIT Lincoln Laboratory and a dedicated team of teaching assistants, provided crucial oversight. Their primary role was to ensure safety, a paramount concern in any project involving high-speed rotating machinery and combustion. Weekly progress reviews served as forums for critical evaluation of student designs and a rigorous assessment of their AI utilization strategies, with Professor Spakovszky employing a nuanced Socratic method to guide teams without directly providing solutions. His pointed queries, such as "Do you know what a rabbet fit is? Take in the comment," encouraged deep reflection and independent problem-solving.
The Four-Week Sprint: A Chronology of Discovery
The challenge unfolded over a highly condensed four-week period, a timeline that itself underscored the potential for AI to accelerate engineering cycles.
Week 1: Conceptualization and Initial Design
By the close of the first week, one team, overwhelmed by the sheer scope and complexity, regrettably withdrew. The remaining six teams, however, had, with varying degrees of completeness and success, developed initial conceptual designs for their gas turbines. This phase saw a robust application of AI tools. Students leveraged LLMs to summarize dense technical textbooks on turbomachinery, rapidly acquire proficiency in commercial design software, identify potential manufacturing vendors, generate complex Excel spreadsheets for calculations, and answer specific, esoteric engineering questions. AI was also instrumental in sourcing relevant academic references and performing comparative analyses between competing design decisions. One particularly innovative team deployed an AI agent within the Parley platform, effectively tasking it with serving as their dedicated project manager, a testament to the students’ ingenuity in offloading organizational overhead. This initial phase demonstrated AI’s undeniable power as a knowledge accelerator and a formidable assistant in the early stages of design exploration.
Week 2: Detailed Design and Encountering AI’s Limitations
The second week marked a critical transition as teams moved from conceptual sketches to detailed CAD designs, initiated the arduous process of ordering bespoke parts, and began prototyping their combustors. It was during this phase that the inherent limitations of current AI models began to surface, challenging the students’ initial enthusiasm. While LLMs like Claude and ChatGPT proved adept at offering alternative design configurations and filling gaps in theoretical knowledge, their notorious tendencies for "hallucinations" (generating plausible but factually incorrect information), "sycophancy" (agreeing with user prompts even when incorrect), and a fundamental "lack of physical understanding" started to erode student confidence and, paradoxically, decelerate their progress.
Elizabeth Tupaj, a member of the victorious 811 Crew, articulated this emerging skepticism: "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 sentiment highlights a crucial distinction: AI excels at information synthesis and pattern recognition, but struggles with the nuanced, context-dependent judgment required for novel engineering design, especially when grounded in complex physical principles. Teaching assistant John Zhang observed this dynamic firsthand, noting, "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 prompt engineering and understanding AI’s inherent biases and limitations from the outset.
Final Weeks: Fabrication, Assembly, and the Unyielding Realities of Manufacturing
As the challenge progressed into its final weeks, the remaining teams encountered an obstacle that no AI could readily solve: the complexities of working with external manufacturing vendors. Students reported significant frustrations: "AI searches found vendors we had no rapport with, who had no interest in our tight timeline." The critical insight gained was that "The vendors who came through were the ones our team had personal relationships with." This experience starkly illuminated manufacturing and supply chain management as a fundamental rate-limiting step, a bottleneck that current AI tools, despite their informational prowess, could not effectively circumvent. The human element of trust, established relationships, and persuasive communication proved indispensable in navigating the logistical labyrinth of procuring custom-fabricated components under extreme time pressure.
Of the three finalists, only the team "Fast and Fractured" achieved first-attempt ignition of their mini-combustor, a significant milestone. This team, despite having no prior gas turbine experience, had heavily leveraged AI for trade studies and architectural comparisons, arriving at a viable design. Professor Masha Folk, the Charles Stark Draper Career Development Professor of Aeronautics and Astronautics, reflected on this 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."
The Final Tally: Victories and Lessons Learned
By the end of May, following the intense four-week sprint, the two most senior teams, "Fast and Fractured" and "811 Crew," successfully completed full engine tests. "Fast and Fractured," despite their AI-assisted design, faced persistent vendor delays, which ultimately cut short their hot fire test when the rotor rubbed and seized against the stationary housing. Team "811 Crew," however, emerged as the ultimate victors. With greater prior exposure to turbomachinery and propulsion concepts, their engine started flawlessly, transitioned successfully to Jet-A fuel, and generated net thrust, fulfilling the challenge’s core objective. PhD student Joe Chiapperi vividly recounted the tension and triumph: "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."
Intriguingly, the "811 Crew" had demonstrated a marked resistance to extensive AI usage throughout the competition, preferring to rely on their foundational knowledge and synergistic 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 younger students, less encumbered by established practices, tended to use Parley more frequently and creatively, the juniors and seniors often leaned on their deeper experiential knowledge.
Professor Andreea Bobu provided a nuanced interpretation of these outcomes: "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 JARVIS Challenge’s clearest and most profound finding was the undeniable multiplier effect of engineering experience and the enduring vitality of the human factor. Mastery of first principles, fundamental concepts, and a robust understanding of physics breeds superior engineering judgment — an invaluable asset for navigating the myriad of tough decisions that arise in the face of incomplete or ambiguous information. When it comes to designing and building safety-critical physical systems, the irreplaceable elements remain human hands, human accountability, and human intuition. 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."
Shaping the Future: Implications for AI-Native Engineering
The implications of the JARVIS Challenge for the future of aerospace engineering and beyond are substantial. If small teams, empowered by intelligently managed AI copilots, can condense design-build-test cycles from years or even months to mere weeks, the ramifications for workforce structures, research and development timelines, and global competitive dynamics could be revolutionary. This paradigm shift suggests a future where smaller, highly agile engineering teams, augmented by AI, could achieve what previously required massive organizational structures.
However, the challenge also highlighted that the current limitations of AI, particularly in areas requiring true physical intuition, common sense reasoning, and the ability to innovate beyond existing data, mean that human engineers are not merely supervisors but essential drivers of the design process. They are the ones who must bridge the gap between AI-generated ideas and the tangible realities of manufacturing, material science, and operational reliability.
Professor Cordero, associate director of the MIT Gas Turbine Laboratory, emphasized the enduring importance of traditional education in this AI-accelerated era: "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 underscores that AI is a tool, albeit a powerful one, and its effective utilization depends on a well-educated, critically thinking human operator. The students who participated in the JARVIS Challenge are at the vanguard of this new era, grappling with these stakes not as theoretical thought experiments, but with actual jet engines on a test stand, forging the path for what it means to be an "AI-native engineer."
