Generative Design
Software that generates many design options against stated goals and constraints.
Quick Answer
Generative design is a design method in which software generates many candidate solutions from a set of goals, constraints, and rules, then evaluates them against measurable criteria. Designers define the problem and select from the results. It lets teams explore far more options than manual drawing, trading time spent drawing for time spent framing the problem.
The Full Picture
In a conventional process, a designer sketches a few options and refines the most promising. In generative design, the designer instead describes the problem: the site boundary, the program, the rules to follow, and the outcomes that matter, such as daylight, floor area efficiency, or structural weight. A program then produces a large set of valid designs and scores each one.
The core mechanism is a loop of generation and evaluation. A generator creates variations by changing inputs within allowed ranges, an evaluator measures each result against the stated goals, and a selection step keeps the better candidates. Some tools use evolutionary or optimization algorithms to improve candidates over repeated rounds, while others simply sample the design space and rank results.
Applications in the built environment include floor plan and massing exploration, structural member sizing, facade patterns, and layout of repetitive elements. The method is closely related to parametric and computational design, but it adds automated evaluation and selection, rather than just allowing a designer to adjust parameters by hand.
The limits are practical. The quality of results depends on how well the goals and constraints capture the real problem, and anything left out of the model, such as local code nuances, tenant preferences, or constructability, is invisible to the algorithm. Outputs are best treated as informed options for a designer to judge, not finished designs.
Real Examples
Common Misconceptions
People assume: Generative design replaces the designer.
Actually: The designer defines the goals, constraints, and evaluation criteria, and judges the results. The tool expands the number of options considered and does not decide what matters.
People assume: Generative design always produces the optimal building.
Actually: Results are only optimal for the criteria that were modeled. Real projects involve factors that are hard to quantify, so outputs typically need human review and refinement.
Frequently Asked Questions
How does generative design work?
The user defines goals, constraints, and parameters. Software then generates many design variations, evaluates each against the goals, and ranks or evolves them so the designer can choose among strong candidates.
What is the difference between generative and parametric design?
Parametric design lets a designer change inputs to update a model by rule. Generative design adds automated generation and evaluation of many variations toward stated goals.
Is generative design the same as AI?
Not necessarily. Many generative design tools use rule-based or optimization algorithms rather than machine learning, though AI techniques are increasingly used to generate or evaluate options.
What are typical uses in architecture and construction?
Floor plan and massing exploration, facade layouts, structural optimization, and site layout studies, particularly where there are many competing measurable goals.