Google’s Gemini Is Not Alone: Adobe’s Firefly Also Trips Up On Race When Generating AI Art

What Happened: Last month, Google temporarily paused its Gemini AI’s capacity to produce images of individuals due to inaccuracies found in the depiction of historical figures. This included depicting America’s Founding Fathers as Black.

Now, Firefly, which uses similar techniques for creating images from written text, was found to replicate many of the same issues that plagued Gemini. Despite being trained on different datasets, both services have faced criticism for their inaccurate racial and ethnic depictions.

Benzinga’s independent testing yielded mixed results, highlighting the ongoing challenge for AI to accurately represent diverse identities.

In response to the backlash, Adobe told Semafor, “Adobe Firefly is built to help people ideate, create and build upon their natural creativity,” adding, “It isn’t meant for generating photorealistic depictions of real or historical events.

“Adobe’s commitment to responsible innovation includes training our AI models on diverse datasets to ensure we’re producing commercially safe results that don’t perpetuate harmful stereotypes,” the company said. “Given the nature of Gen AI and the amount of data it gets trained on, it isn’t always going to be correct, and we recognize that these Firefly images are inadvertently off base.”

Adobe did not immediately respond to Benzinga’s request for comments.

Why It Matters: The controversy surrounding Gemini and now Firefly highlights the challenges that tech companies face in developing AI tools that accurately depict historical events and avoid perpetuating harmful stereotypes.