{"slug": "dragoncatcher-the-secret-of-lightness", "title": "Dragoncatcher: The secret of lightness", "summary": "Robin Sloan argues that language models are fundamentally different from brains and that scaling laws alone may not lead to superintelligence, suggesting that smaller, more efficient models could achieve high performance. He challenges the assumption that massive parameter counts are necessary, advocating for a focus on lightness and efficiency in AI development.", "body_md": "[The secret of lightness](/lab/secret-of-lightness/)\n\nWe do not yet understand how to train language models! This seems obvious to me, because it ought to be possible —\n\nThe famous [Scaling Laws](https://arxiv.org/pdf/2001.08361?utm_source=Robin_Sloan_sent_me) only describe transformer models —\n\nA fair objection goes like this: *Robin, remember that the human brain has hundreds of trillions of “parameters”, in the form of synapses. Our largest models haven’t even approached that scale yet. Do you want us to architect a beetle’s brain, or SUPERINTELLIGENCE?*\n\n(Before proceeding, Robin replies: well, I wouldn’t mind starting with the beetle … )\n\nThe obvious response to this objection is that language models aren’t brains. Contra the brain, they operate with both handicaps (e.g. power consumption) and advantages (e.g. speed). More than linearly “better” or “worse”, though, they are just different! And so we should expect different properties, different capabilities … different numbers.\n\nHanging over all this, the recognition: the day that this level of intelligence moves out to the edge —\n\nA true believer in the Scaling Laws doesn’t think such an idea can exist — [I’m with Calvino](https://zerogravity.empac.rpi.edu/lightness/?utm_source=Robin_Sloan_sent_me):\n\nWere I to choose an auspicious image for the new millennium, I would choose [ … ] the sudden agile leap of the poet-philosopher who raises himself above the weight of the world, showing that with all his gravity he has the secret of lightness, and that what many consider to be the vitality of the times —\n\nnoisy, aggressive, revving and roaring — belongs to the realm of death, like a cemetery for rusty old cars.\n\nOf course, this is just a post by a child of the 20th century, to whom the prefix “giga-“ still sounds unspeakably plush. Even so: if you tell me you can’t fit a supercapable model, one poised comfortably on today’s performance frontier, into 30 billion parameters, I will tell you, try harder!\n\n[To the blog home page](/lab/)", "url": "https://wpnews.pro/news/dragoncatcher-the-secret-of-lightness", "canonical_source": "https://www.robinsloan.com/lab/secret-of-lightness/", "published_at": "2026-06-17 10:00:00+00:00", "updated_at": "2026-06-17 16:30:02.417614+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-ethics"], "entities": ["Robin Sloan", "Calvino"], "alternates": {"html": "https://wpnews.pro/news/dragoncatcher-the-secret-of-lightness", "markdown": "https://wpnews.pro/news/dragoncatcher-the-secret-of-lightness.md", "text": "https://wpnews.pro/news/dragoncatcher-the-secret-of-lightness.txt", "jsonld": "https://wpnews.pro/news/dragoncatcher-the-secret-of-lightness.jsonld"}}