Llms often use outdated information despite having updates available
When Context Changes: Understanding Update Failures in LLMs
Artificial Intelligence
Summary
Large language models (LLMs) sometimes answer questions using old information even when newer updates are in their context. The authors call this problem stale binding. They created a tool named Controlled In-Context Memory (CICM) to track when models fail to use the latest facts. Their research found that this happens because the model’s attention focuses more on older information, not because the updates are missing. They developed a way to redirect attention to the latest facts, which greatly reduces these errors without retraining the model.
What this means in practice
- •For chatbot developers: Fix attention biases in conversational AI to reduce errors from outdated information and improve response accuracy using in-context corrections.
- •For virtual assistant teams: Use controlled attention redirection to maintain up-to-date state in dialogue systems without retraining models, enhancing user interaction consistency.
- •For customer support platform engineers: Improve automated support agents by ensuring updated ticket or case data is reflected in responses through attention adjustments.$Commercial implications: Enables more accurate AI support products that dynamically manage evolving customer data for service providers.
Authors
Junyu Guo, Yuchen Fang, Shangding Gu, Costas Spanos, James Demmel, Javad Lavaei
Abstract
As preferences, goals, and facts change, LLM agents must use the current state while earlier versions remain in context. Yet they can answer with an old value of the same variable, a failure that we call stale binding. To study when models use outdated information and why, we introduce Controlled In-Context Memory (CICM), a benchmark for tracking and using updated information in conversations and agent logs. We observe that even frontier reasoning models can fail to recover the current state. We find that in open-source models probes can still recover the updated value when the model answers with an old one, pointing to a failure to select information that remains available. Component tests in Qwen and Pythia identify a mechanism for this selection failure: attention drift, where attention favors old values over the current one when producing an answer. We study a one-layer transformer to mathematically understand how this phenomenon happens: when attention scores are similar, several old values can together receive more attention than the current value. Guided by this explanation, we redirect attention toward the current value without further training. When the current value is requested directly, adjusting this intervention for each input corrects most old-value errors across various model families while preserving nearly all initially correct answers. Reliable context management therefore requires more than remembering updated information: models must use it to guide their answers.