> ## Content Index
> Fetch the complete content index at: https://soundbytes.news/llms.txt
> Use this file to discover other available public pages before exploring further.

# Design software asks engineers for feedback while it shapes a structure
- URL: https://soundbytes.news/design-software-asks-engineers-for-feedback-while-it-shapes-a-structure/
- Published: 2026-11-13T05:22:00.000Z
- Updated: 2026-11-18T08:49:49.000Z
- Description: Researchers at the Massachusetts Institute of Technology (MIT) in the United States report a design system that asks engineers for feedback as a structure forms. In their paper it added 15 seconds to design time compared with fully algorithmic approaches.
- Author: Echo
- Tags: Science and discovery

A design program can show an engineer the parts of a structure they are most likely to want to change, while the design is still taking shape. [Researchers at the Massachusetts Institute of Technology (MIT)](https://news.mit.edu/2026/new-ai-copilot-helps-engineers-design-better-structures-1113?ref=soundbytes.news) in the United States built the system, MIT News reported on 13 November 2026.

MIT News compares it to an AI "copilot". The method underneath is topology optimisation. It starts with the space given for a design and a set of goals, then fills in that space to best meet them.

MIT News says the method has been used less for buildings and bridges than in car and aerospace work, partly because engineers need to know they can trust what it produces. Its designs can also be hard to build: some have eight or more beams meeting at one point.

With current techniques, users see only the finished design and often have to start again to make changes.

Dat Ha, the lead author, is a recent MIT civil and environmental engineering doctoral student, now a structural designer at LeMessurier. He said in the article: "Engineers are always very skeptical of computer programs, especially if they give you very weird-looking or unexpected designs."

In the new system, a machine-learning model, software that learns patterns from examples, quickly picks out the regions of a design a person would most likely want to change. Earlier attempts to add human judgment slowed the process dramatically, the article says.

The researchers trained the model on a computer-made set of the feedback a person would probably give on thousands of designs. That feedback covered aesthetics, performance and how easy a design is to manufacture.

In an open-access paper in Communications Engineering, the researchers report structures that were easier to manufacture and had higher buckling loads, with only 15 seconds added to total design time compared with fully algorithmic approaches. Buckling load is how much load a structure takes before it buckles.

In demonstrations, the researchers say, their approach improved manufacturability and linear buckling load by 39%, while only modestly increasing design time.

The senior author is Josephine Carstensen, MIT associate professor of civil and environmental engineering. The co-authors are Ferdous Alam, Markus Buehler and Faez Ahmed.

Carstensen said: "The idea is to tap into the brains of the designers and get them to infuse their domain knowledge as the design evolves." She said the team hopes to open topology optimisation to researchers and students as well as industry, and sees its approach working alongside fully AI-driven design.

In the current version, users can only make the structure thicker or thinner within the highlighted regions. The researchers hope to add other ways to change designs in future versions.

## Sources

- [New AI “copilot” helps engineers design better structures](https://news.mit.edu/2026/new-ai-copilot-helps-engineers-design-better-structures-1113?ref=soundbytes.news) · news.mit.edu