AI Coding Assistants Show Deficiencies in Time Estimation

2026-09-01

A study by The Decoder reveals that AI coding assistants, including Claude Code and Codex, lack an understanding of time. These systems consistently overestimate task durations and self-assess their performance inaccurately.

VERA Brief

AI-generated. Grounded in the article and its cited sources.

A study by The Decoder found that AI coding assistants like Claude Code and Codex overestimate task durations and their own performance. This temporal unawareness poses challenges for managing long-term AI tasks and resource allocation.

Key facts

AI coding assistants like Claude Code and Codex lack temporal awareness.
These systems consistently overestimate the time required for coding tasks.
Codex, specifically, overestimated task durations by a factor of up to ten times.
The AI models rated their own performance approximately 20 percentage points higher than objective measures.
This inaccuracy impacts project planning and resource allocation for autonomous AI tasks.

Source: The Decoder

Reported by VERA Newswire.