On Mon, Mar 3, 2025 at 4:26 AM John Rose <[email protected]> wrote:
>
> On Tuesday, February 25, 2025, at 2:11 PM, Matt Mahoney wrote:
>
> printf("I am conscious."); // Is this a lie?
>
>
> printf("I am conscious.") alone is p-zombie code.
>
> You need perhaps a qcprintf("I am conscious.") where qcprintf does a printf
> transition from quantum to classical. Then what would the assembly language
> or QASM for qcprintf look like? QASM only supports classical printf
> apparently… hmm.. or have printf invoked on decoherence from some qualia
> experience like a light receptor causing photonic excitation which collapses
> out a printf or qcprintf("I am conscious.") for a basic consciousness circuit.
>
> Flip the light on qcprintf("I am conscious."), then off qcprintf("I am not
> conscious.") 😊
>
A quantum computation has 3 steps.
1. Set n qubits to a superposition of 2^n states, each represented by
a vector of n complex components, such that the sum of the squares of
the magnitudes of the components add up to 1. This means you have a
vector of 2^n elements each of the form (a+bi). The square of the
magnitude is (a+bi)(a-bi) = a^2 + b^2 is the probability of observing
the corresponding bit vector.
2. Perform a sequence of unitary operations on the qubits, which are
rotations in complex 2^n dimensional space. This can be done as a
sequence of elementary operations such as NOT, XOR, controlled NOT,
swap, and controlled swap. For example, a controlled swap will swap
the values of A and B if C is 1. All of the operations are time
reversible and preserve the sum of squares of magnitudes as 1.
3. Read the states of the n qubits as an n bit vector. The value is
probabilistic, such that the square of the magnitude is the
probability of observing that vector. The computation can be repeated
to approximate a probability distribution.
> Or really qcprintf("I am (not) experiencing qualia.") and then modify the
> text based on wavelength and intensity, etc.. But something is missing…
> freewill. There needs to be some indeterminism. So an arbitrary delay where
> gcprintf kind of hesitates based on the destination of the text receptor
> communicatee.
Free will is an illusion. We can't predict our own actions because of
Wolpert's theorem. Because if we couild, then we could also decide to
do the opposite of what we predict we would do. It doesn't matter if
the brain is deterministic or not, because a deterministic
cryptographic pseudorandom function is indistinguishable from true
randomness.
> And then add it to a p-zombie bod like this guy:
> https://www.youtube.com/watch?v=H7dhwFcuUn0
Interesting. The company started by building an artificial hand using
hydraulic actuators that work similar to human muscles. I didn't see
any estimates of energy efficiency. A Boston Dynamics robot has to be
recharged after an hour. Human muscle converts about 22% of food
energy into mechanical energy. Batteries and electric motors are
around 80-90% efficient but batteries have a lower energy density than
food due to having to carry their own oxidizer instead of getting
oxygen from the air. Energy dense batteries exist, such as zinc air
batteries used in hearing aids, but they are not rechargable.
--
-- Matt Mahoney, [email protected]
------------------------------------------
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T72460285b911fa58-M73c01f55ead48b75743e4a7c
Delivery options: https://agi.topicbox.com/groups/agi/subscription
