Godfather of AI Yann LeCun says: Yes, LLMs may be passing Maths Olympiads and bar exams, but they will still fail in …
Former Meta AI scientist and one of the godfathers of modern artificial intelligence (AI), Yann LeCun, has revealed another task in which humans can beat the technology. Speaking at the India AI Impact Summit 2026, he said that large language models, while “incredibly useful,” still fall short on tasks like driving that humans can handle with relative ease. LeCun acknowledged that AI systems can pass the bar exam and perform well in Maths Olympiads but said a more fundamental gap remains, noting that the technology still lags behind real-world intelligence.“We certainly do not have self-driving cars that can teach themselves to drive in 20 hours of practice, like a 17-year-old,” he said, highlighting that “we’re missing something big”. LeCun explained that AI excels at information retrieval and symbolic reasoning but lacks a true understanding of the physical world. He compared large language models to an evolution of the printing press, libraries and the internet. “It’s just a more efficient way to access information,”he added.
What Yann LeCun said about the gap between AI and real-world intelligence
LeCun said that humans and animals learn through observation and interaction, gradually forming “mental models” that help predict outcomes and adjust to new situations.He also stated that, unlike human learning, AI systems are not yet able to function properly in complex, unpredictable environments, which limits the potential of robots and autonomous cars compared to human learning.Speaking about the use of AI in education and development, LeCun described AI as a technology meant to assist, not replace, human capabilities. According to him, AI has the potential to enhance human intelligence and expand access to knowledge, much as the printing press did in the past.He also noted that countries with younger, educated populations, including India and several regions of Africa, may play a larger role in AI innovation if investments are made in skills development and infrastructure.LeCun, a professor at New York University and the executive chairman of AMI Labs, works in the fields of artificial intelligence, machine learning, and robotics and is also an ACM Turing Award laureate.He also recently publicly questioned OpenAI’s claims about achieving artificial general intelligence (AGI) following a social media exchange with OpenAI vice president Sebastien Bubeck. The discussion, which took place on X (formerly Twitter), reflected differing views on research transparency and whether a single company could develop AGI.LeCun criticised OpenAI’s research practices for being conducted largely behind closed doors and rejected the notion that AGI would emerge from one organisation or a single breakthrough, stating that OpenAI does not have a monopoly on innovation.
