Research Study

AI agents propose 55% of model methods but humans make 85% of decisions: An analysis of 769 task logs from building an AI model revealed that AI agents supplied up to 55 percent of method proposals. However, humans still made over 85 percent of the final decisions. Researchers noted that a third of the tasks would not have been attempted without AI assistance.

Artificial Intelligence • The Decoder

AI access drops people's willingness to admit ignorance from 44% to 3%: A study of over 3,000 participants found that having access to AI answers nearly eliminated people's willingness to say "I don't know," dropping it from 44% to 3% in one experiment. Despite the AI being almost always wrong, users felt more confident but were correct only a third as often.

AI research • The Decoder

Study shows self-distillation can degrade multi-turn LLM agent performance: Researchers found that on-policy self-distillation teaches multi-turn LLM agents to act with confidence without the underlying information, sometimes performing worse than untrained base models. To address this, they proposed Privileged Self-Practice, which moves privileged information from the loss to the sampler.

LLM • arXiv