Why Yann LeCun Says ChatGPT Isn’t Even as Smart as a Dog

As artificial intelligence advances rapidly, debates about its potential dangers intensify. Meta’s chief AI scientist Yann LeCun stated that systems such as ChatGPT are far from human-level intelligence and are not even as smart as a dog.

ChatGPT, developed by OpenAI, is based on a large language model trained on massive text data, enabling it to answer user queries in understandable language.

Many tech experts warn that uncontrolled AI could pose societal risks. Elon Musk has called AI one of the greatest future threats to humanity. At the Viva Tech conference in Paris, economist Jacques Attali noted that AI’s impact depends on its use, citing concerns about AI‑driven fossil‑fuel exploitation and weapon development, while also acknowledging its potential benefits for health, education, and culture.

During a panel at Viva Tech, Yann LeCun explained that current models are limited because they are trained only on language, lacking real‑world understanding. He emphasized that humans acquire most knowledge outside of language, which these systems cannot learn.

LeCun added that AI can pass the U.S. bar exam but cannot operate a dishwasher, a task a ten‑year‑old can master in minutes, illustrating the gap between AI capabilities and everyday human skills.

He concluded that AI is far from human intelligence and not even as smart as a dog, and that Meta is working on training AI for video, a challenging endeavor.

LeCun also highlighted that current AI cannot replicate infant perception of physical reality, such as recognizing that floating objects are abnormal, underscoring the lack of common‑sense understanding.

Will robots take over the world?

Attali expressed pessimism about the future, fearing climate disaster, war, and the possibility of robots rebelling against humans.

LeCun argued that more intelligent machines should not be seen as threats but as beneficial tools, envisioning personal AI assistants that act like smarter coworkers.

He stressed that AI systems must be designed to be controllable and fundamentally obedient to humans, rejecting the sci‑fi notion that smarter robots will inevitably dominate.

AI Ethics and Regulation

Attali called for safety guardrails on AI development but questioned who should set the boundaries. AI regulation is a hot topic at Viva Tech, with the EU advancing AI legislation and French officials advocating for global oversight.