@naval @maxhodak_ @naval read this from @SchmidhuberAI I think it's the best
Some users dismissed Schmidhuber's theory linking consciousness to RNN data compression as misleading romanticist vibes that fail to address unbridgeable gaps like p-zombies.
Most Activity
@SchmidhuberAI take on consciousness
TLDR: 1) General problem-solving agent stores entire obs-action history 2) Compression of data via patterns gives features 3) The agent is common pattern in data, so self-feature arises. 4) Activating that feature can be “self-awareness”
@naval @maxhodak_ @naval read this from @SchmidhuberAI I think it's the best

@dionizije_fa @naval @maxhodak_ @SchmidhuberAI Uff, a lot to digest after a first morning coffee. Where is the orginal source?

@dionizije_fa @naval @maxhodak_ @SchmidhuberAI IDN, There is something ontologically different about a living organic substrate.

@gamzeyaavor @dionizije_fa @naval @maxhodak_ @SchmidhuberAI Romanticist vibes are misleading n every other context, so probably here too

@dionizije_fa @naval @maxhodak_ @SchmidhuberAI Answers nothing. Unbridgeable gap between having a self-symbol and having consciousness. P-zombies can model and factor themselves in computations too.

@RudyDrischel @naval @maxhodak_ @SchmidhuberAI Its cited in the img 🙏

@ZbynekDrab @dionizije_fa @naval @maxhodak_ @SchmidhuberAI Yes, the truth can be misleading to some!

I had to use Grok to read that:
“Consciousness = efficient encoding of the self in a data-compressing world-model during problem-solving.”
If the reading is correct, why use “consciousness” when you know this word means a whole lot of other things as well.
People do mean more than that by consciousness. It may be a property in the encoding, but still people mean more than the encoding (ie the values perhaps)