Search For Powder Intelligence by Windspren

Windspren

30 / 4

2nd Jul 2016

29th Sep 2016

The reborn SFPI! Our goal is to develop and identify forms of pseudointelligence and life within TPT in the forms of PSTN bots, computing systems, genome emulators, brain architecture simulations, and anything else we (or TPT AIs we find :D) can think of

sfpi tpte3

Comments

• 

  Schmolendevice 5th Jul 2016
  I'm still looking into ways to implement learning. PROT doesn't seem to be an option or be reliable. What we need is a way to alter the "strength" of ARAY to increase with the amount by which it is sparked... Oh, we could attach TUNG to the synapses. Hmm.
• 

  QuanTech 5th Jul 2016
  i might also abandon my perceptrons. they are not very efficient. they don't even have more than 2 inputs!
• 

  QuanTech 5th Jul 2016
  and i think i've understood your neurons a bit more.... i'll see how to configure them and simply just work with them!
• 

  QuanTech 5th Jul 2016
  maybe 1-bit addition (with learning) please? (oh god we are so lucky to have a neurologist with us!)
• 

  Schmolendevice 5th Jul 2016
  I'm also unsure of the exact cons against my design lest it is simply confusing to use or it is unclear what the size is of just one of my neurons. @Kevino36 Against your argument of "realistic neurons" filling up the whole screen, five of mine can fit into just one of your neurons that you asked us to work with. Perhaps give me a task and I can try to configure them to perform that operation.
• 

  Schmolendevice 5th Jul 2016
  I do certainly specialize in CPU architecture and FILT electronics, but I wouldn't want to end up with a completely non-biological "more advanced than neurons" AI system. The brain would end up existing entirely programmatic in nature stored in the RAM of a FILT CPU.
• 

  QuanTech 5th Jul 2016
  hey kevino36, i think you spawned an entirely new area of technology: self-replication. Now other people are even making FP saves with it! Who knows, maybe someone may find new applications for it.
• 

  QuanTech 5th Jul 2016
  noooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
• 

  QuanTech 5th Jul 2016
  oh noooooooooo..... my perceptron only has maximum 2 inputs..........................
• 

  QuanTech 5th Jul 2016
  actually, i think i'll start with the logical OR operation (with learning). It seems adding control circuitry and planning the neural networks requires some math....