Archive · 30 Jan 2026
Yann LeCun Criticizes AGI, Agentic AI
Yann LeCun, a famous AI expert, recently spoke at Davos about his concerns with artificial intelligence.
He believes that the idea of AGI, or AI that can think like humans, is overrated.
He thinks that making current AI systems bigger or better won't lead to human-level intelligence.
Instead, he says we need a completely new approach.
LeCun also warned about AI assistants that can act on their own, saying they can't predict the results of their actions.
He gave an example of how a 10-year-old can solve a simple task without much training, but AI needs millions of hours of data to learn.
He also talked about how AI systems are too simple and can't understand the real world.
LeCun is worried about a few big companies controlling AI, which could be bad for democracy and diversity.
He thinks AI research should be open and shared, like it used to be, to make progress.
He also mentioned that the best open-source AI models are currently coming from China.
Yann LeCun, a renowned AI expert, criticizes the idea of AGI, stating it's overrated and scaling current AI won't lead to human-level intelligence.
He argues that agentic AI systems, designed to act autonomously, lack the ability to predict the consequences of their actions.
LeCun believes current AI systems operate in a simplified universe and can't understand the complex real world.
He warns about the consolidation of AI control among a few companies, which could be detrimental to democracy and diversity.
LeCun advocates for open-source AI research, stating that the shift towards closed research labs is slowing down progress.
- Who
- Yann LeCun, a professor and AI pioneer
- What
- Criticism of AGI, agentic AI, and the current AI research paradigm
- When
- Recently at Davos
- Where
- Davos
- Why
- To highlight the limitations and dangers of current AI trends
- Yann LeCun
- Professor and AI pioneer
- AGI
- Artificial General Intelligence
- LLMs
- Large Language Models
- Agentic AI
- AI assistants designed to take actions autonomously
- LeCun's Concern
- Consolidation of AI control among a few companies
- LeCun's Solution
- Open-source AI for diversity and progress
- Current AI Research Trend
- Shift towards closed research labs
- Best Open-Source Models
- Currently from China