Yann LeCun warned that AGI is still years away and argued at the India AI Impact Summit 2026 that AI should enhance, not replace, human intelligence.

LeCun, a foundational figure in modern AI research and former chief AI scientist at Meta, used his platform in New Delhi to rein in what he described as “overblown expectations” around AGI, urging the industry to focus on amplifying human intelligence rather than chasing speculative human‑equivalent AI.

“The real revolution is not AI replacing humans, but AI helping humans think better, plan better, and understand the real world,” LeCun said.

Against this turbulent backdrop, LeCun’s keynote offered a technical and philosophical critique of current AI trends. LeCun argued that despite rapid advances in large language models (LLMs) and other machine learning systems, these tools lack the fundamental capacities that define intelligence, as they do not truly reason or predict, key ingredients of human‑like cognition and they cannot build comprehensive “world models” (internal representations of real‑world experience) that would allow them to generalise beyond specific tasks.

“Agentic systems cannot exist without predicting consequences of actions, and LLMs cannot do this. So we need world models,” LeCun explained.

He insisted that training on text alone is fundamentally limited. LeCun compared the information volume in public internet text to the sensory input a child receives in the first years of life, concluding it was insufficient to build human‑like AI.

“We have to build AI that truly understands the world,” LeCun told delegates, framing the challenge not just as one of computation, but of philosophy and perception.

LeCun stated that AI could boost annual productivity by 0.6 percent, with its true impact seen in long-term scientific and medical progress. He noted that the distribution of these benefits depends on political choices, not just technology.

LeCun believes that countries with large young populations, particularly India and nations across Africa, are positioned to lead the next phase of AI innovation. For India, which is already adopting AI at scale, the central challenge is talent development through sustained skilling and re-skilling.

Bill Gates and Jensen Huang bow out

The sudden withdrawal of Bill Gates from his scheduled keynote slot was confirmed by the Gates Foundation just hours before he was set to take the stage. “After careful consideration, and to ensure the focus remains on the AI Summit’s key priorities, Mr. Gates will not be delivering his keynote address,” the foundation said in a statement.

The summit programme was quick to designate Ankur Vora, president of the Gates Foundation’s Africa and India offices, to speak on Gates’ behalf.

Gates’ withdrawal comes amid renewed scrutiny over his name appearing in documents referenced in the recently released Epstein files – a tranche of materials linked to convicted sex offender Jeffrey Epstein that has stirred controversy globally. Gates has denied any wrongdoing and described allegations in the documents as “absolutely absurd and completely false.”

Last week, NVIDIA’s Jensen Huang also backed out of his scheduled appearance due to “unforeseen circumstances.”

LeCun was the first to break away from AI giants

In late 2025, LeCun walked away from his position as Meta’s chief AI scientist after more than a decade – a departure that was widely viewed within the industry as an early signal in a broader wave of research dissent and lack of faith in the direction of AI.

LeCun was increasingly at odds with Meta’s pivot toward commercialising large language models and rapid deployment, as opposed to pursuing foundational research without immediate product pressure.

LeCun openly criticised aspects of Meta’s strategy, accusing leadership of prioritising benchmarks and product cycles over scientific integrity, an approach he said risked eroding trust within research teams.

LeCun’s departure was one of the earliest high-profile exits in what has since become a broader movement of AI researchers leaving major labs, citing concerns over governance, research autonomy, and ethical trade-offs.

Analysts have noted that these exits are not simply about compensation or career moves but reflect deep tensions over AI’s research agenda and societal impact.

Some departing researchers have explicitly warned that companies are moving too fast and downplaying core limitations and risks of AI systems, raising questions around safety, bias, and long‑term consequences of unregulated development.

Related stories

AI godfather blasts Zuckerberg: Yann LeCun breaks his silence

As AI researchers exit, infrastructure investors watch the governance debate

OpenAI rocked by talent exodus amid claims research is being sacrificed for AI advocacy