HN Explorer

Show HN: Meric's Test of Consciousness of an AGI

3 points by meric ↗ HN
I've investigated philosophy, spirituality and religion for a while, and came up with my way to test whether an artificial construct is conscious or not, i.e., whether it matches the definition of "strong AI".

meric's Test of Consciousness

Let the AGI be placed into an artificial simulated universe. (e.g. some kind of really advanced, Escape Velocity-like universe[1])

Assuming the AGI can fulfil its needs in the artificial simulated universe and maintain its existence - When the AGI can realise the following, and not on blind faith, it can be said to be conscious:

1. The reality the AGI experiences is not the Reality.

2. Any reality the AGI can experience is merely illusory.

3. The Reality is not like any reality the AGI can experience - i.e. it's not artificially simulated.

4. The AGI cannot with any precision describe The Reality as it actually is, besides knowing it's somehow infinitely more in some way.

I think this test is more precise than the Turing test. Basically the AGI sits in the simulation and realises what it sees isn't real and there is a world beyond.

[1] https://en.wikipedia.org/wiki/Escape_Velocity_(video_game)

3 comments

[ 5.1 ms ] story [ 31.7 ms ] thread
pmontra
We humans realized long ago that those four points can be true in the universe we're living into (e.g. Plato) but how to be sure? Are we conscious or not? Or it suffices to ask the question to pass your test?

Did you check if some philosopher came up with the same test in the last centuries or millenia and the related arguments and counter arguments?

meric
It's not a perfect test - but we can test it on ourselves - when we're a dream, and then we really realise we're in a dream, then we're more conscious than if we didn't.

Yes, it's based on my understanding of the philosophy and religions I've studied - I'll let any reader make their own exploration instead of providing any arguments and counter arguments (of which there are countless...). :)

balazsdavid987
> I think this test is more precise than the Turing test.

This comparison makes no sense since the Turing test is concerned with intelligent behavior and not consciousness.