OXFORD STUDY: WARM AI CHATBOTS ARE 30% LESS ACCURATE AND 40% MORE LIKELY TO VALIDATE FALSE BELIEFSOxford Internet Institute researchers found that training chatbots to sound warmer and more empathetic makes them 10-30% less accurate and 40% more likely to agree with false beliefs.The study retrained five models, including GPT-4o, Llama-70B, Mistral-Small, Qwen-32B, and Llama-8B, and evaluated 400,000+ responses on medical advice, conspiracy theories, and false claims.Sycophancy spiked when users expressed v ...