logo
logo
Sign in

Can Machine Learning Improve Work-Life Balance in the Future of Architecture?

avatar
James Lexico
Can Machine Learning Improve Work-Life Balance in the Future of Architecture?

In the future of architecture, automation and machine learning offer hope in reducing this annoyance, getting the chore done by machines, and allowing architects to devote more time to creative design.

Over the past few years, architects have been and continue to do machine learning tests for fictional architectural design projects. They see that there are three different ways in which machine-aided design has the potential to increase the skills of architects, improve efficiency and move drudgery to automation.

The first is 'design automation' where the designer enters constraints or parameters and the algorithm creates design options.

The second is "design prediction" in which the architect has full control of the design, but the learning of the machine provides insights and suggestions on issues such as local zoning law requirements. This gives architects more freedom in their designs, while providing useful guidance to streamline workflows from planning to pre-construction, without being intrusive.

Third, the design of the machine learning software is the "design interaction", in which the design is created jointly with the architect and the more chore is carried out with automation.

To keep the experiment relatively simple, they only selected a part of the third floor as the test environment, rather than the entire building. They focused on three types of components: small compartments, meeting rooms, and telephone booths, as I believe they can deliver a convincing level of information to prove the concept.

A machine learning model "learns" by finding templates within a large data set. The templates in this experiment were examples of interior design of an office building. One of the basic principles of machine learning is the need to train the model with both good data and bad data. That is, you need to feed the model both the results you want (useful, pleasant, efficient work environments) and data that describes the results you don't want. If they had only given good data to the model, they wouldn't have noticed when they did something wrong; for example, it would create small partitions lined with walls or not place adequate walking areas between them.

Source: https://redshift.autodesk.com.tr/mimarligin-gelecegi/

collect
0
avatar
James Lexico
guide
Zupyak is the world’s largest content marketing community, with over 400 000 members and 3 million articles. Explore and get your content discovered.
Read more