DigestAI news desk

Cut through the AI noise.

Research1 min read

OpenAI data shows AI automates coding, monitoring over decision-making

OpenAI shared internal data comparing how its researchers used AI tools between late January and early August. The company found AI-assisted output surged in coding and infrastructure tasks—198,200 tokens per researcher per day for research and infrastructure code, 158,800 for technical help and review, and 133,100 for launching, monitoring, and debugging. These tasks have clear, measurable…

1 source

Key points

  • Researchers used AI 198,200 tokens/day for coding and infrastructure tasks in August vs. January
  • Decision-making tasks averaged 2,300 tokens/day for focus decisions, 200 for stopping work
  • OpenAI says AI automates measurable tasks but not judgment or prioritization

Decision-making tasks saw far less automation: only 2,300 tokens per researcher per day for focus decisions, 1,500 for computing and staffing, and 200 for continuing or stopping work. Planning research and experiments generated 5,100 tokens daily. OpenAI notes AI excels at execution but struggles with judgment, prioritization, and compromise—skills still central to research decisions.

Full story from thenews.com.pk · by Pareesa Afreen · via Search: OpenAIOpen source ↗

OpenAI's own data shows AI automates execution, not decisions

thenews.com.pk · 26 September 2026

Loading the full article…

This text was published by thenews.com.pk and written by Pareesa Afreen. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Research

All →

Related stories