Quest to build a strong artificial intelligence
My goal here is to build a real time biomimetic brain. That means that I will try to virtually replicate a real human brain. I hope to create an intelligence while doing it.
Why realtime? I want it to learn something before I die ^^.
Being realtime will force me to do some compromises that I hope won’t prevent intelligence to exists. I will explain all compromises I will do and why. If you disagree on one point fell free to tell me and try your one way!
Lets get a plan on how to proceed.
Plan
Step 1 The senses
Consider the brain as a black box: don’t consider yet how the brain is working. Just analyze what are its inputs and outputs. A better understanding of the nature of its input is compulsory to do biomimetism.
- The retina (redaction ongoing)
- Ears (redaction todo)
- touch (redaction todo)
- musles (redaction todo)
Step 2 Senses modelisation
Build libraries to generate these inputs and outputs.
- retina modelisation (redaction ongoing)
- Ears (redaction todo)
- touch (redaction todo)
- musles (redaction todo)
Step 3 The brain
Dive on how the brain works. How it links all theses inputs and outputs (neurons, cortical areas, hormones, …)
Step 4 Brain modelisation
Implement a brain making some compromises to keep it realtime.
- neurons (redaction on going)
- cortical areas (redaction on going)
- hormones (redaction on going)
- What we can currently see in the brains? (irmf, public databases) (redaction on going)
- experimentations
- …
(redaction on going)
Step 5 Brain expériments
Send inputs to the brain and try to detect signs of intelligence. If I can’t detect it:
- try to find macroscopic differences with a real human brain => go back to step 3 to find why it is not happening in human brain.
- look over all compromises
(redaction on going)
Step 6 Are you sure?
I already went to step 5 and I want to present what I understood on the way. I hope you will learn something ^^