← Back to People
Byline author analyzing AI progress against the AI 2027 scenario, discussing Claude Mythos capabilities and whether explosive AI progress is occurring.
4
Mentions
4
Articles
4
Outlets
Aug 2025 — Oct 2026
Coverage
Negative
Sentiment
Associated Topics
Companies Covered
Coverage Patterns
How media typically covers Steve Newman
Article Types
- analysis1
- opinion1
Media Angles
- technical2
Narrative Framing
- cautionary1
Appears In
Sentiment Distribution
Based on 3 scored articles
Mixed 33%Negative 67%
Associated AI Models
GPT-52GPT-11GPT-41ChatGPT1
Writes About
Helen TonerResearcher
2 articlesJason KwonResearcher
1 articleJack ClarkResearcher
1 articleDean BallPolicy
1 articleRandi WeingartenPolicy
1 articleNicholas CarliniResearcher
1 articleNathan LabenzCreator
1 articleDaniel KokotajloResearcher
1 articleEli LiflandResearcher
1 articleScott AlexanderCreator
1 articleYoshua BengioResearcher
1 articleRo KhannaPolitician
1 articleJan KulveitResearcher
1 articleVivian SchillerExecutive
1 articleEli PariserCreator
1 articleAppears Alongside
Articles
Most recent firstArticles Written
Steve Newman as author
35 Thoughts About AGI and 1 About GPT-5
Second ThoughtsopinioncautiousAug 18, 2025
Current AI systems lack specific missing functions like reliable judgment and continuous learning that distinguish them from AGI, and the path to AGI may be either characterized by breakthrough 'aha' moments or slow incremental grinding.
“Author of "35 Thoughts About AGI and 1 About GPT-5" in Second Thoughts”
AI SafetyFoundation ModelsAI GovernanceLLMs
Quoted In
Directly quoted in these articles
The Extreme Inefficiency of RL for Frontier Models
Toby OrdanalysiscautiousSep 22, 2025
The shift from pre-training to reinforcement learning for frontier models reduces information efficiency by 1,000 to 1,000,000x, potentially constraining future scaling progress despite unlocking new reasoning capabilities.
“Quoted discussing the challenge of training models to independently manage month-long projects with limited learning per unit time”
Foundation ModelsReinforcement LearningAI ResearchLLMs
Also Mentioned In
Referenced in coverage
What I saw around the Curve, 2026
Second Thoughts AIcautiousOct 13, 2026
“Opened The Curve 2026 conference noting rapid developments make it difficult for individuals to understand AI changes.”
Is AI 2027 coming true?
secondthoughts.aicautiousJun 19, 2026
“Byline author analyzing AI progress against the AI 2027 scenario, discussing Claude Mythos capabilities and whether explosive AI progress is occurring.”