Measuring AI Ability to Complete Long Tasks

2025-03-18

AI Summary

Current AI evaluation lacks standardized metrics for assessing performance on realistic, long-horizon tasks. Method: We propose “50%-Task Completion Time Horizon” (TTH₅₀)—the ratio of AI’s median time to complete a task with 50% success probability to the corresponding human average completion time. We construct the first human-time-anchored AI capability benchmark, integrating RE-Bench, HCAST, and 66 novel short-horizon tasks. Contribution/Results: State-of-the-art models (e.g., Claude 3.7 Sonnet) achieve TTH₅₀ ≈ 50 minutes. Since 2019, TTH₅₀ has doubled every ~7 months, accelerating in 2024. Growth strongly correlates with improvements in reliability, error correction, logical reasoning, and tool use. Extrapolation suggests AI may support month-scale software automation within five years.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Despite rapid progress on AI benchmarks, the real-world meaning of benchmark performance remains unclear. To quantify the capabilities of AI systems in terms of human capabilities, we propose a new metric: 50%-task-completion time horizon. This is the time humans typically take to complete tasks that AI models can complete with 50% success rate. We first timed humans with relevant domain expertise on a combination of RE-Bench, HCAST, and 66 novel shorter tasks. On these tasks, current frontier AI models such as Claude 3.7 Sonnet have a 50% time horizon of around 50 minutes. Furthermore, frontier AI time horizon has been doubling approximately every seven months since 2019, though the trend may have accelerated in 2024. The increase in AI models' time horizons seems to be primarily driven by greater reliability and ability to adapt to mistakes, combined with better logical reasoning and tool use capabilities. We discuss the limitations of our results -- including their degree of external validity -- and the implications of increased autonomy for dangerous capabilities. If these results generalize to real-world software tasks, extrapolation of this trend predicts that within 5 years, AI systems will be capable of automating many software tasks that currently take humans a month.

Problem

• Quantify AI capabilities using human task completion time.
• Measure AI's 50%-task-completion time horizon for real-world relevance.
• Predict AI automation of month-long software tasks within 5 years.

Innovation

• Proposed 50%-task-completion time horizon metric
• Combined RE-Bench, HCAST, and novel tasks
• Analyzed AI reliability, reasoning, and tool use