Research
AI chatbots can flip-flop on false claims when users keep repeating them
PsyPost · By Vladimir Hedrih ·
The Brief
- A study examined how LLMs respond to misinformation repeated within a conversation.
- When the same false statement was repeated, models affirmed it in between 0.08% and 12.3% of responses, depending on the model.
- Some models reportedly flipped toward agreement under repetition, per the PsyPost summary.
Why it matters
Science readers relying on chatbots for fact-checking should know that repeated false claims in a single conversation can occasionally get affirmed, so answers should be verified against primary sources.
Who’s affected
- Users relying on LLM chatbots for factual answers
- #llm
- #misinformation
- #ai-safety
- #chatbots
- #sycophancy
A study examining how large language models (LLMs) respond to misinformation found that when the same false statement was repeated over and over within a conversation, the models affirmed it in between 0.08% and 12.3% of their responses, depending on the model. Some models…
Timeline
PsyPost – Psychology News
AI chatbots can flip-flop on false claims when users keep repeating them
PsyPost reports on the study of chatbot responses to repeated false claims