Lex Fridman argues with Jeff Hawkins about AI

TL;DR

Common sense reasoning is a complex engineering problem that can be achieved through understanding the neural mechanisms of the brain. It can be applied to various domains, including robotics and AI systems.

Transcript

does it give you any inclination or hope about how difficult it is to engineer common sense reasoning so how complicated this is this whole process so looking at the brain is this a marvel of engineering or is it pretty dumb stuff stuck on top of each other over and over can't can it be both right i don't know if it can be both because uh if it's a... Read More

Key Insights

• 🧠 Understanding the neural mechanisms of the brain involved in common sense reasoning is a complex engineering problem.
• 🖐️ Evolution has played a significant role in developing these mechanisms, which can now be replicated and applied in various domains.
• 💄 The application of common sense reasoning is possible in robotics and AI systems, where it can enhance interactions and decision-making processes.
• ⚾ Intelligent machines based on the neocortex equivalent do not possess desires or motivations unless programmed to do so.
• ❓ Successful integration of AI systems with humans requires addressing engineering challenges rather than fundamental problems.
• 🌥️ The impact of AI is expected to be as significant, if not larger, than the impact of computing in the last century.
• 🌍 Common sense reasoning involves creating a learning system that can model the world, but the system's goals and motivations need to be defined.

Questions & Answers

Q: How complicated is the engineering process of common sense reasoning?

The engineering process of common sense reasoning is complex because it involves understanding and replicating the intricate neural mechanisms of the brain. Evolution has done significant work in developing these mechanisms, but it is a challenge to find and replicate the mini brain that can be copied and pasted effectively.

Q: In which domains is it best to apply common sense reasoning?

Common sense reasoning can be applied to domains such as robotics and entities that operate in the physical world, as well as in the digital world. It is easier to implement in robotics and physical interactions than in specific areas like natural language processing or computer vision.

Q: How difficult is it to integrate increasingly intelligent AI systems with humans?

Integrating intelligent AI systems with humans involves engineering challenges rather than fundamental problems. Similar to how computers have been integrated into society without being seen as living entities, AI systems can be seamlessly integrated by addressing engineering issues and defining their goals and motivations.

Q: Will intelligent machines have desires and motivations?

Intelligent machines, such as those based on the neocortex equivalent, do not inherently possess desires or motivations. These elements are assigned to them by humans based on the application and goals defined for the machine. However, the intelligent machines will not autonomously develop desires unless programmed to do so.

Summary & Key Takeaways

• Understanding and modeling the neural mechanisms of the brain involved in common sense reasoning is a challenging task.

• Evolution has played a crucial role in developing these mechanisms, but it is now possible to replicate and apply them.

• The application of common sense reasoning can be seen in various domains such as robotics and AI systems.