A decade in, public AI literacy still fails: researcher calls LLMs 'stochastic parrots'
gerardsans · x · 2026-10-06
Responding to John Nosta, researcher gerardsans laments that nearly a decade later, the public and even researchers publish work that fails basic AI literacy, calling the amount of AI misinformation "staggering."
His blunt framing of LLMs: text in, text out — in the middle, a probabilistic function throwing words back at you. "Not a ghost. More like a stochastic parrot. A text sampler, if you want to sound cooler."
More from AGI Musings
- AI differs from past automation because it automates automation itself — burny_tech · 2026-10-06
- Veteran dev mourns dying coding craft: from hand-sliced Photoshop corners to AI — CtrlAltDwayne · 2026-10-06
- AI Race Is Now a Five-Layer Contest: Intelligence, Compute, Energy, Manufacturing, Speed — ingliguori · 2026-10-06
- LLMs expanded the space of trivial results — publish fewer papers and call out paper mills — bremen79 · 2026-10-06
- Burry vs. AI bulls: critic argues AI has a virtuous loop dot-coms never had — robleclerc · 2026-10-06
- AI Ethicist Carissa Veliz's 'Prophecy' Shortlisted for FT Best Business Book Award — CarissaVeliz · 2026-10-06