In other news from Google's annual I/O developer conference, the tech giant showcased a variety of Gemini's other features, including custom chatbots, a vacation itinerary planner, and integrations with Google Calendar, Keep, and YouTube Music. However, the image generation feature remains switched off in Gemini apps on both web and mobile platforms. The ongoing issue serves as a reminder of the challenges in addressing bias in AI technology.
Key takeaways:
- Google's AI-powered chatbot Gemini's ability to generate images of people has been paused since February due to complaints of historical inaccuracies and bias in its image generation.
- Despite promises of a fix, the issue has not been resolved as of May, with the image generation feature still switched off in Gemini apps on the web and mobile.
- The problem stems from the data sets used to train image generators like Gemini, which contain more images of white people than people of other races and ethnicities, and often reinforce negative stereotypes.
- Google is struggling to find a middle path that corrects these biases without resorting to clumsy hardcoding, highlighting the complexity of addressing bias in AI.