MIT Undergraduates Build 50-100-Pound-Thrust Jet Engines in 4 Weeks With AI
Updated
Updated · MIT News · Jul 14
MIT Undergraduates Build 50-100-Pound-Thrust Jet Engines in 4 Weeks With AI
1 articles · Updated · MIT News · Jul 14
Summary
31 MIT students in seven teams spent four weeks designing, fabricating and testing small jet engines in the JARVIS Challenge, with 811 Crew ultimately producing net thrust after transitioning its engine to Jet-A fuel.
AI sped early work such as textbook summaries, trade studies, software help and architecture comparisons, but students said hallucinations, weak physical intuition and sycophantic answers often slowed detailed design decisions.
Manufacturing proved the main bottleneck: vendor delays and fabrication constraints outlasted design work, and Fast and Fractured’s AI-assisted engine test ended when its rotor rubbed and seized.
The clearest result was that engineering judgment and prior experience mattered more than AI use alone, with senior-heavy teams generally outperforming younger students even as the latter used AI more aggressively.
MIT faculty said the sprint suggests AI copilots could compress aerospace design-build-test cycles from years to weeks, but only when guided by strong first-principles training and hands-on engineering education.
AI can design a jet engine in weeks, but can it fix the real bottleneck: manufacturing and supply chains?
As AI automates design, how do we train engineers to develop the critical judgment that machines still lack?
If AI lacks physical intuition, what breakthrough is needed for it to safely design real-world, critical hardware?
From Years to Weeks: The JARVIS Challenge and the Future of AI-Accelerated Engineering at MIT
Overview
The JARVIS Challenge at MIT in early 2026 brought artificial intelligence out of theory and into real-world engineering by making AI the main partner for undergraduate teams building small jet engines. This hands-on program aimed to show if AI copilots could help small teams shrink the design-build-test process from years to just weeks. Students directly experienced both the power and limits of AI in engineering, revealing how AI can speed up complex projects but still needs human judgment and expertise. The challenge highlighted a new era where engineers and AI work side by side to accelerate innovation.