⁉ What is AGI

General intelligence does not have one well-defined meaning, and it does not mean the same thing to all researchers. However, Artificial General Intelligence has a clear qualitative meaning to the term; the ability to solve complex new problems with a high degree of understanding autonomously. Human-level understanding and problem-solving.

🤺 AI vs AGI

A bulk of AI today has been concerned with “narrow AI”, solving very specific problems like identifying a disease or predicting the next word in a sentence, but AGI has hopes of becoming a general problem solver. A program built-in with common sense and the ability to generalize cross-domain.

🤔 Imitating all of the human intelligence?

In order to achieve technology on par with human-level intelligence this requires replicating human mental processes. Including thought, memory, attention, decision-making, language and more. But this leads to the question, what about negative human features? Should AGI also mimic undesirables features like bias? From here on we will focus on the preferred approach, AGI with the absence of negative features.

💻 Current Approaches

AGI is an exciting field in which numerous theories and approaches are currently being tested out. Here are some of the most promising approaches being tested out today.

Algorithmic probability

• Prediction of the environment around the agent using Solomonoffs induction and the Bayes theorem. Based on predictions the agent will choose the action maximizing future reward; the future is described as an agent's lifetime.

Integrative cognitive architecture

• Identify the central cognitive processes that occur within the human brain and then replicating them individually within an AGI.

There are 3 popular forms of cognitive architecture