Can Artificial Intelligence Design an Aircraft Engine? The MIT Experiment

The JARVIS Competition, held at the Massachusetts Institute of Technology, examined whether generative AI tools could design, test, and build a small gas turbine aircraft engine within four weeks.

Purpose and Scope of the Experiment

The Massachusetts Institute of Technology Gas Turbine Laboratory launched the JARVIS Competition to see the role of artificial intelligence in the physical product development cycle. Participating teams were asked to quickly design and manufacture a small aircraft engine running on Jet-A kerosene.

Use of Artificial Intelligence

In addition to traditional engineering programs, the teams utilized a new AI interface called MIT Parley, which includes large language models like ChatGPT and Claude. The system significantly accelerated the process in conceptual studies and literature reviews.

Limitations Encountered by Artificial Intelligence

During the detailed CAD design and prototyping phase in the second week, artificial intelligence reached its limits due to hallucinations and a lack of fundamental physics understanding. Supply chain and physical manufacturing processes were also among the bottlenecks that the algorithms could not overcome.

The Role of Engineering Expertise

As a result of the competition, it was observed that the team relying most heavily on artificial intelligence made rapid progress in the initial phase but failed the final hot-fire test. First place was won by the team with traditional expertise applying fundamental engineering knowledge.