{"slug": "world-models", "title": "World Models", "summary": "A deep learning practitioner argues that training world models on sanitized data is counterproductive, as exposing models to unfiltered reality at scale is necessary for unexpected capabilities to emerge.", "body_md": "A world model trained on a sanitized world is a contradiction. My deep learning career taught me: at low data, curate carefully. At scale, stop protecting models from reality. That is where unexpected capabilities emerge. https://t.co/1fO6faqNZV", "url": "https://wpnews.pro/news/world-models", "canonical_source": "https://twitter.com/selimonder/status/2078540842822733993", "published_at": "2026-07-18 19:39:24+00:00", "updated_at": "2026-07-18 19:51:13.936143+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "artificial-intelligence"], "entities": [], "alternates": {"html": "https://wpnews.pro/news/world-models", "markdown": "https://wpnews.pro/news/world-models.md", "text": "https://wpnews.pro/news/world-models.txt", "jsonld": "https://wpnews.pro/news/world-models.jsonld"}}