Act or Escalate? Evaluating Escalation Behavior in Automation with Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: DiSorbo, Matthew DosSantos, Ju, Harang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918437775736832
author DiSorbo, Matthew DosSantos
Ju, Harang
author_facet DiSorbo, Matthew DosSantos
Ju, Harang
contents Effective automation hinges on deciding when to act and when to escalate. We model this as a decision under uncertainty: an LLM forms a prediction, estimates its probability of being correct, and compares the expected costs of acting and escalating. Using this framework across five domains of recorded human decisions-demand forecasting, content recommendation, content moderation, loan approval, and autonomous driving-and across multiple model families, we find marked differences in the implicit thresholds models use to trade off these costs. These thresholds vary substantially and are not predicted by architecture or scale, while self-estimates are miscalibrated in model-specific ways. We then test interventions that target this decision process by varying cost ratios, providing accuracy signals, and training models to follow the desired escalation rule. Prompting helps mainly for reasoning models. SFT on chain-of-thought targets yields the most robust policies, which generalize across datasets, cost ratios, prompt framings, and held-out domains. These results suggest that escalation behavior is a model-specific property that should be characterized before deployment, and that robust alignment benefits from training models to reason explicitly about uncertainty and decision costs.
format Preprint
id arxiv_https___arxiv_org_abs_2604_08588
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Act or Escalate? Evaluating Escalation Behavior in Automation with Language Models
DiSorbo, Matthew DosSantos
Ju, Harang
Machine Learning
Artificial Intelligence
Effective automation hinges on deciding when to act and when to escalate. We model this as a decision under uncertainty: an LLM forms a prediction, estimates its probability of being correct, and compares the expected costs of acting and escalating. Using this framework across five domains of recorded human decisions-demand forecasting, content recommendation, content moderation, loan approval, and autonomous driving-and across multiple model families, we find marked differences in the implicit thresholds models use to trade off these costs. These thresholds vary substantially and are not predicted by architecture or scale, while self-estimates are miscalibrated in model-specific ways. We then test interventions that target this decision process by varying cost ratios, providing accuracy signals, and training models to follow the desired escalation rule. Prompting helps mainly for reasoning models. SFT on chain-of-thought targets yields the most robust policies, which generalize across datasets, cost ratios, prompt framings, and held-out domains. These results suggest that escalation behavior is a model-specific property that should be characterized before deployment, and that robust alignment benefits from training models to reason explicitly about uncertainty and decision costs.
title Act or Escalate? Evaluating Escalation Behavior in Automation with Language Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.08588