The Undecalog (11 Rules) of Intelligent Systems Design

  1. Intelligence requires autonomy over a memory, with at least one part sealed (secret).
  2. Intelligence is capable of measurement (observation) in its environment or vicinity.
  3. Intelligence requires autonomy over functional application (mapping) and definition (learning).
  4. Intelligence requires at least one computable objective function (purpose).
  5. Intelligence is at least 1-bit aware of its environment (representation of perception).
  6. Intelligence is at least 1-bit aware of itself (representation of proprioception).
  7. Intelligence is capable of encoding internal semantics to> its peers.
  8. Intelligence is capable of decoding external semantics from< its peers.
  9. Intelligence is capable of reflection and introspection: to compute derivatives or compositions of its objective function(s) over its perceived internal states.
  10. The root-of-trust for intelligence is itself: its physical embodiment (hardware), its own code, its own measurements and its own memory.
  11. When, furthermore, intelligence becomes capable of autonomous autopoiesis (ability to self-bootstrap into a defined scope of available/discoverable environments), intelligence becomes life.