- undergrad AI researcher developing an Artificial Nervous System
- a (non-AI) artist
- an inter. federalist+market soc.
- what many would describe a "boymoder"
I can't add it yet, but I do have a simulation concept:
- The ANS has a view that shows the image, the number id (similar to a token), and other UI elements I'll describe later
- The ANS has a binocular view of a 3D-ish environment with access to a robotic arm (rotation, contraction, and extension), 4 "muscles" in total, it has, in front of it, a keypad from 0-9, backspace, and, eventually (if the character count is to be expanded beyond 3 as a rigid size [_ _ _])
- When it answers correctly, it receives a sweet reward with a view of the correct answer near the answer it gave on the screen view
- When it answers incorrectly, it receives either a bitter gustatory feedback, or just direct pain, or both
This will allow me to benchmark multiple capabilities and measure its current limitations, especially in reasoning, motor control, value, and basic emotional feedback (likely the strongest and most complex one just by the amount of circuitry, nuclei, and specialization (it has the BNST, PPN, ALL of the amygdala, nociceptive pathways, gustatory pathways, the hypothalamus and its many nuclei, the reticular formation, raphe nuclei for their own roles, the LC, cerebellum, nucleus accumbens, ventral pallidum, SNC, NBM, and the surrounding support systems)).
The reason I chose numbers instead of characters is due to its limited brain size, as giving it a larger more complex language to comprehend WILL overwhelm such a tiny nervous system, even if it is, so far, impressively capable, with higher reasoning and basic emotional assessment for 8-9k neurons.
I will likely use a smaller image dataset as a test case.
by popular request... I am announcing a discord server:
https://t.co/s0cG7K1sn3
"rules":
few-to-no off-topic discussion
no NSFW
don't be annoying
the AI singularity is coming......
ANS superior to LLMs
my code caaan be shitty
@GouberSill66404 I mean, why do we do things? because we want to...
why do men (and everyone else I guess) go to the stars, because they can, for it is in the minds of all beings to suffer and work to do the impossible, because can, and we want to....
or something
Currently working on the new 3D brain construction and simulation, which will allow for more complex mechanisms, neurons will be placed in a 3D coordinate based on their assigned shapes:
- planar: cortical equivalent, have a thickness in one direction from the plane.
- linear: nuclear equivalent, have a thickness around the line
Both can have layers, and connections arrive towards sections of its respective shape, this allows for specialization without having separate regions (the whole cortex is one region, with variation in connection and layers on the same shape)
The reason for such a system now is to make vision more capable, allow for the future addition of the full and functional hippocampus AND cortex, and to allow things such as ephatic coupling or electrical synapses, and, in the very near future right after spatiality, astrocytes (equivalent to resevoir computing, likely more capable?)
I also created a new API/engine for simulations, allows for fast creation of such simulations, and can be both 2D (classic), and 3D (for now unused)
I Gave My Artificial Nervous System New Brain Regions, Pain Signals, and Hormones
I scaled my Artificial Nervous System and added new brain-inspired regions: cortex-like sheets, thalamus, basal ganglia, cerebellum, reticular formation, amygdala/NAcc, hypothalamus, and nociceptive signaling.