Browsing: Bias

Recent advances in AI interpretability are shedding light on how large language models work, revealing both their capabilities and potential biases.

Critics argue that the popular Chatbot Arena benchmark for ranking AI models is skewed in favor of large technology companies

Research from Cohere, Stanford, MIT, and Ai2 alleges that Chatbot Arena allowed select AI companies to privately test models and selectively publish results

Meta’s latest AI model, Llama 4, is under fire for suggesting conversion therapy, a discredited practice condemned by medical organizations. Critics accuse Meta of bias and attempting to appease right-wing views.

As generative AI becomes more prevalent, it’s crucial to understand and mitigate potential biases and ensure fairness to unlock the technology’s benefits.