Search papers, labs, and topics across Lattice.
This study evaluates the effectiveness of three families of uncertainty quantification (UQ) methods鈥攚hite-box scorers, black-box consistency scorers, and reflexive scorers鈥攚hen applied to interactive trajectories generated by LLM agents. The findings reveal that while some transfer of UQ methods from single-turn outputs to multi-turn interactions is beneficial, it is inconsistent, with reflexive scores often serving as the most reliable low-cost baseline. Notably, black-box self-consistency methods outperform others in many scenarios, emphasizing the need for revalidation of UQ techniques at the trajectory level, considering factors like aggregator choice and computational efficiency.
Reflexive scores often outperform traditional UQ methods in multi-turn interactions, revealing a critical gap in how we assess uncertainty in LLM agents.
Uncertainty quantification (UQ) methods for language models are typically evaluated on single-turn outputs, where uncertainty is attached to one generated answer. For LLM agents, however, the unit of observation is an interactive trajectory, where the model can ask clarifying questions, call tools, update state, and make intermediate decisions whose errors propagate to the final outcome. We study whether three common families of single-turn UQ methods transfer to this setting. Across five LLMs and four multi-turn tool-use datasets from BFCL-v4 and $\tau^2$-bench, we evaluate white-box scorers based on action-token probabilities, black-box consistency scorers based on resampled trajectories, and reflexive scorers based on model self-assessment of the trajectory. We find that transfer is often useful but uneven. Token-probability scores are highly sensitive to the choice of aggregator used across turns, reflexive scores provide the strongest low-cost baseline in most evaluated settings, and black-box self-consistency is often the strongest UQ family, with trajectory-equivalence and action-set consistency typically ranking highest among its variants. These results suggest that UQ methods developed for single generations should be revalidated at the trajectory level, with careful attention to the consistency measurement, aggregator choice, and computational budget.
