Nanowire 'brain' network learns and remembers 'on the fly'
phys.org
For the first time, a physical neural network has successfully been shown to learn and remember "on the fly," in a way inspired by and similar to how the brain's neurons work.
  • LughOPMAEnglish
    7·
    1 year ago

    It’s fascinating how similar to the brain’s structure this is and how it also mimics its functions. Current AI models built on silicon chips are often assumed to lead to human-level AI - maybe that might be more likely to come from something like this? It also makes me wonder if this might be the right approach for interfacing with the brain for brain implants.

  • peopleproblems@lemmy.worldEnglish
    2·
    1 year ago

    Now what would be really cool is if they take this idea, and add wires with different chemical properties to the mix. Then have certain volumes with different configurations of the different wires, using the same algorithm.

    If that produces results where the configured location correlates between multiples of these devices, we could be looking at a possible implementation of application specific hardware. Companies wouldn’t need to rent supercomputers, they’d buy these “Application Specific Brain On a Chip” devices, train them, then hook em up. Energy usage would plummet, and processing time would skyrocket.

    This is really, really cool.