Abstract

In human-centered design (HCD) projects, designers select and use a variety of design methods in pursuit of a desired outcome. Given the prominence of method selection in designer behavior, what distinguishes a design team’s method selections from design method selection based on frequency or probability? To explore this question, we compare HCD methods suggested by the publicly-available large-language model, GPT-3.5, to 402 novice design team method selections over five offerings of a design projectbased learning course at a large public university. We observe that GPT-3.5 appears to represent design method knowledge held in method repositories like theDesignExchange well. We also observe that GPT-3.5’s method selection recommendations appear to poorly distinguish between HCD phases, and appear limited to highly specific aspects of HCD phases. These findings highlight the unique contribution of human design cognition in design decision-making relative to LLM’s, and herald the promise of human-AI teaming in design method selection.

Keywords

design theory and methodology; human-ai collaboration; human-centered design; design methods

Creative Commons License

Creative Commons Attribution-NonCommercial 4.0 International License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

Conference Track

Research Paper

Share

COinS
 
Jun 23rd, 9:00 AM Jun 28th, 5:00 PM

Exploring human-centered design method selection strategies with large language models

In human-centered design (HCD) projects, designers select and use a variety of design methods in pursuit of a desired outcome. Given the prominence of method selection in designer behavior, what distinguishes a design team’s method selections from design method selection based on frequency or probability? To explore this question, we compare HCD methods suggested by the publicly-available large-language model, GPT-3.5, to 402 novice design team method selections over five offerings of a design projectbased learning course at a large public university. We observe that GPT-3.5 appears to represent design method knowledge held in method repositories like theDesignExchange well. We also observe that GPT-3.5’s method selection recommendations appear to poorly distinguish between HCD phases, and appear limited to highly specific aspects of HCD phases. These findings highlight the unique contribution of human design cognition in design decision-making relative to LLM’s, and herald the promise of human-AI teaming in design method selection.

 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.