Iterated creation, grading, and revision of data science projects by language models

Summary

A study evaluated agentic AI models on end-to-end data science projects with iterative feedback. Across 147 initial submissions from seven models, average scores rose from 59% to 76% after three rounds, with gains varying by model (3 to 46 percentage points). Self-grading inflated scores by about 9 percentage points, and half of round-to-round differences were within grader noise.

This is a reviewed summary. Read the complete article at the original source.

Why it matters by industry

• Artificial intelligence

  Relevant to AI models and agents: quantifies iterative feedback effects and self-grading bias in agentic workflows.