Josh Vermillion, a friend and colleague at UNLV, has been exploring a new tool that uses artificial intelligence to generate images based on verbal prompts. Although I am largely ignorant to how the software operates my basic understanding is as follows: Based on verbal prompts the software searches a database for relevant content and morphs various items into a composite graphic — developing a unique image that didn’t exist prior to the prompt. The entire process takes a few minutes, and if you don’t like the results you can ask for another round of images to be generated or simply modify your prompt. The images Josh has shared are powerful and voluminous. Due to the speed and quantity of images that can be generated the maker is largely orchestrating the parameters of the prompt, curating and selecting images for further development. The image shown here is the result of a three-minute architectural exploration.
For those of us who have spent countless hours and thousands of dollars developing similar images or renderings of proposed buildings, this tool is cause for great excitement and extreme concern. Granted, these images are currently two dimensional and would require an enormous effort to translate into a viable building proposal. However, as an ideation tool it appears to be remarkable. Not only does the program appear to have access to more information than the human mind can retain, but one of its hidden advantages is the balance the maker has between control and chance. If a maker controls everything in a process, the result will always be limited to what they have seen or experienced before. It is only when they let go and allow for the unexpected to occur that true innovation can emerge. An important feature of making is to calibrate the relationship between one’s control and the unplanned, happy accident.