Qualcomm has used AI to generate an image in under 15 seconds using their their second generation Snapdragon 8 mobile chipset, which contains a Hexagon processor designed for AI. They claim it is the fastest they’ve ever generated an image using Stable Diffusion on a phone.
Stable Diffusion is a machine learning model that converts descriptive text into an image. For example: You can type something like ‘blue bee’, and it would generate an image of a blue bee.
The resolution of the image Qualcomm generated is 512 by 512 pixels. That isn’t a high resolution. However, it’s a start, and it’s also an improvement considering that image generation can take much longer on older Snapdragon chipsets. The implications of AI-powered image generation are significant, considering that it can generate detailed artwork in seconds — something that humans take hours to do.
Aside from the employment implications for artists and graphics designers, there are also stock photo companies that make money by selling other peoples’ artwork. Some companies (for example, Shutterstock) have chosen to embrace AI image generation and deploy their own generation tools, but not all of them are happy about it considering that image generation apps will negate the need for these companies eventually.