DeepMind’s New Approach to Avoiding Hallucinations in Large Language Models
DeepMind’s innovative method to reduce hallucinations in Large Language Models (LLMs) by enabling these models to self-evaluate their responses and abstain from answering when uncertain. This approach, known as conformal abstention, enhances the reliability and trustworthiness of AI systems like OpenAI’s Chat-GPT and Google’s Gemini by ensuring they provide accurate and consistent information.