Budget Brilliance, or Cheaper Elsewhere? Google’s Pixel 8A Packs a Punch with Next-Level AI Enhancements! Would You Buy It?
May 17, 2024Google’s Secret Plan Unveiled: Your Phone Could Be Monitoring Your Calls!
May 17, 2024Google’s Struggle with Gemini’s Biassed Image Generator
Google’s AI chatbot Gemini remains hampered by its image generation feature, disabled since February amid concerns of biassed depictions. Despite assurances from Google executives, including CEO Sundar Pichai and DeepMind co-founder Demis Hassabis, the issue persists well into May. Instances like depicting “Zulu warriors” as stereotypically Black highlight the ongoing challenges in rectifying biases ingrained in AI datasets.
The Complex Challenge of Bias Correction
The delay in resolving Gemini’s image generation problem underscores the complexity of addressing biases deeply embedded in AI datasets. With datasets often containing more images of white people than those of other races, attempts to correct biases can lead to clumsy solutions. Google’s struggle to find a balanced approach reflects the broader challenge of navigating AI misbehaviour rooted in systemic biases.
Navigating AI Misbehaviour Amid Systemic Bias: But How Long Must It Take? Competition Is Intense!
As Google continues to grapple with finding a solution, other competitors are creating and releasing faster, more efficient and even more precise than Gemini Image Generator. How long must Google tinker with Gemini before it gets left behind in the dust? Imagine, in a space of a few months, over ten new image generators and image editors have been released. Time is ticking in the AI world, and it waits for no man.
Source: TechCrunch
*An AI tool was used to add an extra layer to the editing process for this story.