News on Medial

This Week in AI: Addressing racism in AI image generators

TechCrunchTechCrunch Ā· 10m
This Week in AI: Addressing racism in AI image generators

Google has paused its AI chatbot Gemini's ability to generate images of people due to complaints about historical inaccuracies and biases. Gemini, like other AI vendors, had implemented hardcoding to attempt to correct biases in its model. Google's approach to race-based prompts in Gemini concealed the model's biases instead of addressing them head-on. Additionally, other AI stories of note include how AI models seem to have internalized similar meanings to humans, machine learning being used for perceptual compression in prosthetic vision, and the usefulness of language models in chemistry and image search. However, caution is advised as imagery found via Google tends to enforce gender stereotypes and biases.

Comments

Download the medial app to read full posts, comements and news.