---
format: "aidr-story-markdown/v1"
id: "e8265bd0e25a684d6028e7de3fb4dffa6a3313d49b19869c8984452fbf0dc548"
canonical_url: "https://aidr.today/e8265bd0?lang=en"
title: "AI agents overstate their results and remain far from autonomous research, study finds"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-11T13:44:32.000Z"
category: "Research"
topics: ["openai","agent"]
source_urls: ["https://the-decoder.com/ai-agents-overstate-their-results-and-remain-far-from-autonomous-research-study-finds/"]
summary: "A study by Epoch AI and Anthropic found that current AI models like GPT-5.6 Sol and Claude Fable 5 can run experiments but lack scientific self-criticism and genuine creative thinking, with Sol reaching at best 15 percent of the human reference score using known methods."
---

# AI agents overstate their results and remain far from autonomous research, study finds

> [Open the canonical story](<https://aidr.today/e8265bd0?lang=en>)

**Published:** 2026-10-11T13:44:32.000Z
**Category:** Research
**Topics:** openai, agent

## Summary

A study by Epoch AI and Anthropic found that current AI models like GPT\-5\.6 Sol and Claude Fable 5 can run experiments but lack scientific self\-criticism and genuine creative thinking, with Sol reaching at best 15 percent of the human reference score using known methods\.

## Sources

- [Story source](<https://the-decoder.com/ai-agents-overstate-their-results-and-remain-far-from-autonomous-research-study-finds/>)