Google’s image generation tool, part of their Gemini AI suite, is set for a swift relaunch after being pulled due to inaccuracies and questionable outputs. Google DeepMind CEO Demis Hassabis acknowledged the issues, aiming for a “fixed” version in a few weeks.
The tool generated racially diverse images when asked for historically specific depictions, leading to accusations of either being too “woke” or lacking proper AI ethics safeguards. This comes on the heels of Google’s rebranding of their Bard chatbot, now also named Gemini.
The issues highlight ongoing tensions in the AI industry. Some criticize a lack of nuanced ethical considerations, while others see a bias towards political correctness. This reignites the debate sparked by Google’s handling of its AI ethics group and the “rushed” rollout of Bard.
Speaking at the MWC24, Hassabis addressed the debacle and said the company had intended for Gemini to show a “universal range of possibilities” in the images it produces, accounting for the many different countries where Gemini is available.
