AGI won’t use computations

🌈 Abstract

The article discusses the concept of Patom theory, which proposes that the brain operates through pattern matching rather than computation. It argues that this approach could be a key step towards achieving Artificial General Intelligence (AGI). The article explores the limitations of the computational model of the brain and how Patom theory offers a more effective way to emulate human brain function.

🙋 Q&A

[01] Patom Theory and Brain Function

1. What is the key idea behind Patom theory? Patom theory proposes that the brain operates through automatic pattern matching, rather than computation. It suggests that brain regions store unique patterns and match them again, which could automate brain function.

2. How does Patom theory differ from the computational model of the brain? Patom theory argues that the brain does not need to send around encoded information or execute instructions like a digital computer. Instead, it stores and matches patterns where they are found, without the need for duplication, search, and data corruption.

3. What are the advantages of Patom theory over the computational model? Patom theory claims it can solve or explain previously unsolvable problems, such as parsing for Natural Language Processing (NLP), by using pattern-matching instead of computation. It also suggests that Patom-based systems can be more versatile and efficient than computational approaches.

[02] Patom Theory and AGI Development

1. How does Patom theory relate to the development of Artificial General Intelligence (AGI)? The article argues that Patom theory could be a key step towards achieving AGI, as it provides a way to emulate human-like capabilities, such as language and complex motion control, through pattern-matching rather than computation.

2. What are the current limitations in AI that Patom theory aims to address? The article suggests that the current limitations in AI, such as the inability of Large Language Models (LLMs) to drive cars or generate human-level text, could be overcome by adopting a pattern-matching approach like Patom theory.

3. How does Patom theory's approach to Natural Language Understanding (NLU) differ from other approaches? The article states that Patom theory's pattern-matching approach to NLU is more effective than previous, expensive and unsuccessful approaches, as it can manipulate meaning with complexity and scale.

[03] Comparison of Brains and Computers

1. How do brains differ from digital computers in terms of data processing and storage? The article highlights that brains do not need to encode, copy, and execute instructions like digital computers. Instead, brains store and match patterns in a fixed location, avoiding the need for data duplication and search.

2. What are the key differences between the computational model and Patom theory's approach to brain function? The article explains that the computational model requires the brain to "plan the sequence of muscle contractions needed" for tasks like throwing a ball, while Patom theory models this as a sequence of patterns that the brain learns and selects through pattern-matching.

3. How does Patom theory's approach to visual perception differ from a computational approach? The article uses the Kanizsa's triangle example to illustrate how Patom theory's pattern-matching can easily explain the brain's ability to recognize shapes that are not physically present, in contrast to a computational approach that would require complex programming to "figure out" the shapes.