Wong Pei Artificial General Intelligence a gental introduction

Table of Contents

This article particularly talked about:

follow is some reading notes and compact reference I draw from the original

1. time

GPT-4(2023) is included, so Dr.Wong’s have updated the tutorial fairly recent. However, the bulk of the tutorial dates before 2014 (before 2023, 2014 is the last year mentioned in the tutorial), so there might be a gap of time to filled in.

  • it is possible nothing important happend during the gap
  • it is possible they are filled in the literature part and is not yet very conclusive

2. approaches of AGI

There’s several ways to define intelligence, and corresponding to each there’s approach to replicate differenct abstraction/model of human consciousness

structure
simulate brain
behaviour
simulate human behaviour
capability
general problem solver
function
simulate each cognitive function(perceive, reason, learning, communication, problem sovling…)
principle
principle guided agent

3. strategies to engineer AGI

hybrid
individual function -> combine
integrated
architecture -> modules
unified
core -> increment

4. list of probable techinques used

  • logic
  • probability theory
  • production system
  • graph theory
  • knowledge base
  • learning algorithm
  • neural network
  • evolutionary computation
  • robotics
  • multi-agent system

Backlinks

old index

(References > AGI)

artificial general intelligence

Artificial General Intelligence is field particularly discussing the general aspect of AI, i.e. general problem solver, artificial consciousenss and such, with awareness of narrow AI.

There are various categorizations attempts, and in particular, there are basically 2 definitions of AGI:

  • Performant fuzzy reasoning agent, with ability to synthesize one or more knowledge bases (database, training dataset or network weights) and perform a variety of tasks, e.g. a robot that can derive subtasks when you tell it to “get me a cup of tea” or “do my taxes” and perform them with good performance. An agent that eat everything and can do everything
  • Developmental agent, with abitility to produce knowledge and develop itself.
  • agi-intro is one up-to-date first material for the field
  • I guess this field is not yet very integrated. there are a lot of disconnected key papers

Here’s a script to insert multiple org-roam nodes 

(defun hermanhel-strings-to-hash (strings)
  "Convert a list of STRINGS to a hash table with the strings as keys."
  (let ((hash (make-hash-table :test 'equal)))
    (dolist (str strings)
      (puthash str t hash))
    hash))

(defun hermanhel-org-roam-insert-multiple-nodes-as-list ()
  (interactive)
(let
    (
     (candidates (hermanhel-strings-to-hash (org-roam--get-titles)))
     (selected-nodes (citar--select-multiple "References: " candidates))
     )
(dolist (title selected-nodes)
      (insert "+ " "[[roam:" title "]]" "\n")
      )
)
)

Backlinks

old index

(References > AGI)

artificial general intelligence

Artificial General Intelligence is field particularly discussing the general aspect of AI, i.e. general problem solver, artificial consciousenss and such, with awareness of narrow AI.

There are various categorizations attempts, and in particular, there are basically 2 definitions of AGI:

  • Performant fuzzy reasoning agent, with ability to synthesize one or more knowledge bases (database, training dataset or network weights) and perform a variety of tasks, e.g. a robot that can derive subtasks when you tell it to “get me a cup of tea” or “do my taxes” and perform them with good performance. An agent that eat everything and can do everything
  • Developmental agent, with abitility to produce knowledge and develop itself.
  • agi-intro is one up-to-date first material for the field
  • I guess this field is not yet very integrated. there are a lot of disconnected key papers

Here’s a script to insert multiple org-roam nodes 

(defun hermanhel-strings-to-hash (strings)
  "Convert a list of STRINGS to a hash table with the strings as keys."
  (let ((hash (make-hash-table :test 'equal)))
    (dolist (str strings)
      (puthash str t hash))
    hash))

(defun hermanhel-org-roam-insert-multiple-nodes-as-list ()
  (interactive)
(let
    (
     (candidates (hermanhel-strings-to-hash (org-roam--get-titles)))
     (selected-nodes (citar--select-multiple "References: " candidates))
     )
(dolist (title selected-nodes)
      (insert "+ " "[[roam:" title "]]" "\n")
      )
)
)

Author: Linfeng He

Created: 2024-04-03 Wed 19:37