Date:
March 07, 2024
Figma’s AI tool Make Designs gets pulled after users complain that it generates near-replicas of Apple products.
Figma is an amazing tool that supports over 4 million users trying to build something unique. The tech giant recently introduced an AI feature called Make Designs, which did not fare well with consumers. Those who’ve used it claim that the Generative AI tool created near-similar replicas of Apple’s weather app. As a result, the company is now temporarily disabling the AI feature till further notice.
Make Designs is a feature within the Figma work system that generates UI layouts and other components based on text prompts. Andy Allen, the founder of NotBoring Software, tried the Figma Make Designs AI tool to create a weather app. The final result looked almost identical to the Apple weather app. He took it to X and shared his feedback, which fueled the controversial claims regarding the nature of Figma’s AI generations.
Figma CEO Dylan Field took the blame for speeding up the development of the Make Designs AI tool over a tight deadline, which may have caused inefficient performance. The company, however, claims that it majorly relies on third-party AI languages, and the team is exploring the extent to which the design similarities come from there. According to Figma CTO Kris Rasmussen, the features are “powered by off-the-shelf models and a bespoke design system that we commissioned, which appears to be the underlying issue.”
Figma has declared that it does not use consumer data to train its AI models. The company has also clarified that it will not use any consumer data for training AI models until it introduces a transparent policy that gives complete control to users about the content-sharing decision. The company has updated a notice in its AI approach blog, where August 15th is set as the official day of launching the new settings. No content training will occur before that, and only the users who agree to share their content for AI training will be considered.
Figma considers texts, images, comments, annotations, layer names, and properties as content that can be used for AI training purposes. It also segregated Usage data as distinct from user content, which includes:
- Information about how your organization’s content is used (like how many times it is accessed)
- Technical logs and metadata
- Telemetry data
After 15th August onwards, the default setting for individual users will be set to ‘ON’, and they will be prompted to switch it off manually. For organizations and Enterprise accounts, the default setting will be ‘OFF’.
