Yann LeCun, a prominent figure in artificial intelligence, has stated that human-level AI, often referred to as artificial general intelligence (AGI), could be achieved within the next 5 to 10 years. This timeline aligns closely with estimates from other AI leaders, including Sam Altman and Demis Hassabis. LeCun emphasized that advancements in AI will not be realized merely through scaling existing large language models (LLMs), which currently lack reasoning and understanding capabilities. Instead, he advocates for a shift towards developing 'Advanced Machine Intelligence' (AMI), a program being pursued by Meta, which aims to address the limitations of current AI technologies. LeCun's remarks have sparked discussions about the potential implications of achieving AGI, including the need for societal readiness and the coordination of institutions to manage the transition. Some experts have expressed skepticism regarding the timeline, suggesting that the pace of innovation largely depends on the focus of researchers on architectural improvements rather than merely scaling existing models.
Yann LeCun announced that Meta is developing an "Advanced Machine Intelligence (AMI)" program, emphasizing that current LLMs are limited by their lack of reasoning, memory, and understanding of the physical world. https://t.co/y5Y5gXjGqT
We are close to achieving AGI by 2025. OpenAI models, o1 and 4o, are nearly there. We just need it to be multimodal.
Meta is working on advanced machine intelligence. They prefer it over LLM https://t.co/FM8UWpc8Bn