AI Trends

Trending story

Jakub Pachocki Advocates Using Perplexity to Establish Common Safety Standards

1 reports1 story updatesUpdated Nov 20 08:00

Understand this story

AI overview

On November 22, 2026, OpenAI Chief Scientist Jakub Pachocki advocated using perplexity, or cross-entropy loss, as a metric for comparing model developers and establishing common safety standards, as well as publicly disclosing some internal pretraining data. The materials show that GPT‐3.5 and GPT‐4 have log-perplexity scores of approximately 0.99 and 0.91, respectively, on major evaluation datasets, with new capabilities in cybersecurity and mathematics emerging below 0.8. He also proposed that developers establish common development limits based on measurements such as perplexity and refrain from crossing them until specific milestones in understanding and monitoring AI behavior have been reached. There are currently no other developments supported by factual evidence.

AI-generated from public reports

Latest development · Report

OpenAI Chief Scientist Jakub Pachocki on Perplexity and Intelligence

Developments

1 confirmed developments

  1. Jakub Pachocki advocates using perplexity to establish common safety standards

    Source:English text is being prepared.

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Nov 20

  1. Trending on Hacker NewsSelected

    OpenAI Chief Scientist Jakub Pachocki on Perplexity and Intelligence

    OpenAI Chief Scientist Jakub Pachocki argues that perplexity or cross-entropy loss should be used as a metric for comparing model developers and establishing shared safety standards, and has disclosed some internal pretraining data. GPT‑3.5 and GPT‑4 have log-perplexities of approximately 0.99 and 0.91, respectively, on their primary evaluation datasets; once the figure fell below 0.8, new capabilities emerged in cybersecurity and mathematics. The article proposes that developers establish shared development thresholds based on measurements such as perplexity and refrain from crossing them until specific milestones in understanding and monitoring AI behavior have been reached.

Related stories

  1. Related storiesMusk Reveals Tesla Plans to Release a Real-World Optimus Demo Within Two Months
  2. Related storiesOpenAI Shares New Data on How AI Is Changing Its Internal Research Methods
  3. Related storiesTencent Hunyuan releases Hy4 flash weights
  4. Related storiesGemini 4 Neon Begins Gradually Rolling Out on the Gemini and Antigravity Platforms