I am half way through building an chemical process and assembly line to etch and fabricate SMARTYPANTS processors (my own design inspired by memristors, quantum physics, genetic algorithms and neural networks). I have just finished programming a ‘helping hands’ robot to be able to be taught motions by copying my actions. But I haven’t finished working out the etching process quite yet, should take me a couple of weeks to sort it all out. The plans is to build a form of AI, using these processors, that can help me out with my projects, an exocortex.
After messing around with a few different brain-computer interfaces (EEG, optogenetics, imaging …) I think I have found a viable way to monitor the full neural state. Now the question is how to decode it and figure out a better input to the brain (higher bandwidth with internet connectivity). Hmm, I think I need to learn some more math for this higher dimensional analysis, let’s go!
So after building a electron tunelling genome sequencer a year ago, I have finally got enough data (from cells across my body) to map my genome and epigenome. I am just trying to figure out the best way to process the data. I have also been running experiments of human tissue cultures in vitro, with my genome/epigenome inserted, and be observing their behaviour under cases of interest, different genes added/removed/methylated, or different environments.
After tinkering with a few different structures for optimising information processing and information representation I am starting to get a good understanding on what does and doesn’t work. I currently have an optimiser that can solve any model in O(n*log(n)) time and O(n^2) space complexity. I recently tested the system in an AI competition where you had to control the actuators on a robotic bird, inspired by humming birds, to achieve certain flight patterns. I came 9th. I think there a few places I could make some big gains, I reckon i’ll do better at the next one.
Today we (I have been collaboration with a bunch of hacker type biologists and programmers on the net) just finished helping building an open source software compiler for a programmable bio-molecule synthesiser and assembler. Effectively we have allowed people to code biology on a computer, which translates the computer commands into DNA, folds of proteins and chains of amino acids.
A couple of computer games I published a few weeks ago (the first was a relativistic game of golf and the other was a quantum racing game) have started to get some attention. I got a couple of comments saying they were pretty good, and someone even asked me if they wanted to collaborate. Anyway, they have started to bring in a little cash which is nice.
I have been in contact with a few different researchers bouncing ideas of them and we think we have found something. Proff. Jack O’Dare wants us (me and a few others in the conversations) to write a paper together related to how we can increase the integration of information using multiple optimisers in a system with a universal neural network. Made from nodes with multiple means of communication and multiple states.
Dreams can come true.