Yann LeCun raises $1B to build AI that understands the physical world (wired.com)

Article summary

Yann LeCun, Meta's former chief AI scientist, has raised $1 billion to fund his new startup, Advanced Machine Intelligence (AMI), which aims to develop AI world models that understand the physical world. LeCun believes that current large language models (LLMs) are limited and will not lead to human-level intelligence. AMI plans to build open-source technology and work with companies in various industries to optimize their operations. The startup's goal is to create a new breed of AI systems that can reason, plan, and interact with the physical world.

Main themes

• AI world models
• LLMs and their limitations
• Human-level intelligence
• Impact of AI on society
• Continual learning and backpropagation
• Formal logic and human reasoning
• Attentional autonomy and learning corpus

What commenters say

• Some commenters believe that LLMs are limited and that world models are necessary for achieving human-level intelligence, while others argue that LLMs can be improved to achieve this goal.
• The development of AI raises concerns about its potential impact on society, with some arguing that it will only benefit the rich and powerful, while others believe that it can bring benefits to all of humanity.
• There is a debate about the fundamental bottleneck to achieving AGI, with some arguing that it is continual learning and backpropagation, while others believe that it is the lack of a world model.
• Some commenters think that current AI models are not intelligent in the way humans are, as they cannot foresee or anticipate events that are unlikely or non-existent in their training data.
• The idea of using how real neurons learn as training data to learn how to learn in the same way is proposed as a potential solution to the limitations of current AI models.
• There is a discussion about the role of formal logic in human reasoning, with some arguing that humans are not naturally logical and that their reasoning is often based on pattern matching and associations.
• The importance of attentional autonomy and the ability to steer one's own learning corpus is highlighted as a key aspect of human intelligence that is currently lacking in AI models.