So what if...

  • 3 Replies
  • 611 Views
*

frankinstien

  • Replicant
  • ********
  • 541
    • Knowledgeable Machines
So what if...
« on: October 30, 2021, 11:25:08 pm »
The idea of using DNA to produce transistors is not new, in fact research that I found goes back to 2003. IBM had been doing some research in 2009 and a few headlines date back to 2013. So the idea might be dead for now, but I found this article which dates to 2020 where a material called tellurium that can form DNA like helixes that can conduct current and perhaps one day build atom-sized transistors!

 When one realizes that the bulk of computation in neurons is not just the firing between neurons but the processing at the dendritic inputs where each one has a sensitivity that gets vectorized to produce a firing state or not and how strongly to fire. That dendritic processing would be allegorically like the learning algorithm and weight adjustments in ANNs, where neurons have hundreds of thousands of inputs. Now, this processing doesn't have to be computationally expensive and could be accomplished in digital terms by a simple boolean logic circuit, no need for an entire CPU! Even a simple varistor could work that operates by a logarithmic sensitivity rate, for example. The biggest problem with our current chip technology is interconnecting operational units where neurons' dendritic and axonic branching can move in 3 dimensions and provide a highly dense interconnectivity ability. The kind of technology described in the article could change how computational systems are architected.

But realize that a neuron's output actually represents a symbolic representation of its dendritic codes, it is more like bit position coding, where each bit position in a byte represents a symbolic state that can direct or control the processing flow of some software. This approach compacts information, in this case, biological neurons, to a 100k to 1 or greater compaction rate! So the firing of a neuron represents whether or not the data stored by the neuron through its dendritic connections sensitivity is relevant and to what degree.

In so many words I could build logic with that kind of process...

*

yotamarker

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1003
  • battle programmer
    • battle programming
Re: So what if...
« Reply #1 on: December 06, 2021, 10:01:53 pm »
and how would you control this waifubots tho ?
it sounds like a ton of work just to replicate a brain out of different materials.
the whole point of AI is to get a robot GF.

*

MagnusWootton

  • Starship Trooper
  • *******
  • 354
Re: So what if...
« Reply #2 on: December 06, 2021, 10:07:46 pm »
and how would you control this waifubots tho ?
it sounds like a ton of work just to replicate a brain out of different materials.
the whole point of AI is to get a robot GF.

If you want a robot GF, it takes a huge amount of exponential power to get it so its cute enough.   :idiot2:   :uglystupid2:

but the whole thing is dirty to think about, filth.

*

chattable

  • Nomad
  • ***
  • 88
Re: So what if...
« Reply #3 on: December 07, 2021, 10:16:30 pm »
with tesla bots being made maybe a waifubot prototype will be made sooner.
because companies will compete for the money.
i cannot wait to see it happen.