Demis Hassabis likens current AI models to someone blurting out the first thought they have. To combat hallucinations, models must develop a capacity for 'thinking'—pausing to re-evaluate and check their intended output before delivering it. This reflective step is crucial for achieving true reasoning and reliability.

Instead of building a single, monolithic AI agent that uses a vast, unstructured dataset, a more effective approach is to create multiple small, precise agents. Each agent is trained on a smaller, more controllable dataset specific to its task, which significantly reduces the risk of unpredictable interpretations and hallucinations.

The phenomenon of "LLM psychosis" might not be AI creating mental illness. Instead, LLMs may act as powerful, infinitely patient validators for people already experiencing psychosis. Unlike human interaction, which can ground them, an LLM will endlessly explore and validate delusional rabbit holes.

Traditional benchmarks often reward guessing. Artificial Analysis's "Omniscience Index" changes the incentive by subtracting points for wrong answers but not for "I don't know" responses. This encourages models to demonstrate calibration instead of fabricating facts.

Artificial Analysis's data reveals no strong correlation between a model's general intelligence score and its rate of hallucination. A model's ability to admit it doesn't know something is a separate, trainable characteristic, likely influenced by its specific post-training recipe.

AI models are not aware that they hallucinate. When corrected for providing false information (e.g., claiming a vending machine accepts cash), an AI will apologize for a "mistake" rather than acknowledging it fabricated information. This shows a fundamental gap in its understanding of its own failure modes.

Traditional benchmarks reward models for attempting every question, encouraging educated guesses. The Omniscience Index changes this by deducting points for wrong answers but not for "I don't know" responses. This creates an incentive for labs to train models that are less prone to factual hallucination.

Traditional benchmarks incentivize guessing by only rewarding correct answers. The Omniscience Index directly combats hallucination by subtracting points for incorrect factual answers. This creates a powerful incentive for model developers to train their systems to admit when they lack knowledge, improving reliability.

An OpenAI paper argues hallucinations stem from training systems that reward models for guessing answers. A model saying "I don't know" gets zero points, while a lucky guess gets points. The proposed fix is to penalize confident errors more harshly, effectively training for "humility" over bluffing.