AGI From First Principles

Over the years, there have been numerous attempts to pin down the essence of AGI, Artificial General Intelligence. Let's break down AGI by first principles, engineering it from the molecular level up—consider it the DNA of intelligence, complex yet elegantly simple in its ultimate design. The focal points for developing such an elegant and complex system include, but are not limited to, the following starting points:

Focus on Capabilities, Not Processes:

AGI should be judged by what it can do, not how it does it. This means AGI doesn't need to think or understand like humans, nor does it need consciousness or feelings.

Focus on Generality and Performance:

AGI should be both general (able to do many different tasks) and perform well in those tasks.

Focus on Cognitive and Metacognitive Tasks:

AGI should be able to perform mental tasks and have self-awareness skills like learning new tasks or knowing when to ask for help. Physical tasks (like moving objects) are helpful but not necessary for AGI.

Focus on Potential, Not Deployment:

Whether AGI can perform certain tasks is more important than actually using it in real-life scenarios.

Focus on Ecological Validity:

The tasks used to measure AGI's progress should be meaningful and practical, reflecting real-world skills people value.

Focus on the Journey not the End Goal.

AGI should be seen as a series of levels, each with specific abilities and benchmarks, rather than a single goal.