Researchers are questioning Google's claims about its AI agents building an operating system for under $1,000. They argue that the "single prompt" description is misleading, as the prompt was thousands of lines long and the process involved a specialized scaffold and agent oversight. Furthermore, Google has not provided evidence to show the agents wrote the code from scratch rather than copying existing material, nor have they released the prompt, code, or logs for independent verification. AI
IMPACT Raises questions about the reliability of AI agent capabilities in complex software development and the transparency of company demonstrations.
RANK_REASON The article is an opinion piece by researchers critiquing a company's product claims, rather than a direct product release or research finding.
- Andrew Schwartz
- Antigravity 2.0
- Arvind Narayanan
- Gemini 3.5 Flash
- Rishi Bommasani
- Sayash Kapoor
- Stephan Rabanser
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
