The short version

Yann LeCun predicted on April 26, 2023: “On the highway towards Human-Level AI, Large Language Model is an off-ramp.” Verdict: Partially right. LeCun argued LLMs alone won't reach AGI and pushed JEPA/world-models. By 2026 frontier labs increasingly added world-models, video, and reasoning beyond pure next-token prediction — partial vindication. Pure-LLM scaling kept producing gains he downplayed.

Prediction by

Yann LeCun

On April 26, 2023

“On the highway towards Human-Level AI, Large Language Model is an off-ramp.”

Follow-up

LeCun argued LLMs alone won't reach AGI and pushed JEPA/world-models. By 2026 frontier labs increasingly added world-models, video, and reasoning beyond pure next-token prediction — partial vindication. Pure-LLM scaling kept producing gains he downplayed.

From the High Caliber AI network — see the AI for PR module in the course.