ºÎ»ê½Ãû µµ¼­¿ä¾à
   ¹Ìµð¾î ºê¸®Çνº³»¼­Àç´ã±â 

åǥÁö





  •  

    Thinking Currents, Sensing Circuits
    - The Next-Generation Artificial Neural Network Inspired by the Human Brain

    Artificial intelligence has long imitated the act of ¡°thinking.¡± Yet circuits that truly 'feel' and 'respond' like the brain have remained beyond reach. In 2025, a groundbreaking study published in 'Nature' crossed that boundary once again. A new device, dubbed the 'trans-neuron', can dynamically switch between different biological neuron functions—sensory, motor, and cognitive—within a single artificial element. It is the birth of a 'multimodal spiking artificial neuron', a technology that transfers the brain¡¯s adaptive complexity onto silicon.


    Circuits That Resemble Neurons — Electronic Implementation of Multiple Senses
    The essence of this research lies in 'multimodality'. Conventional artificial neurons are fixed to one task: receiving sensory input, transmitting motor output, or mimicking cognition. The trans-neuron, however, can 'switch its role depending on the surrounding stimulus'.

    By integrating semiconductor-based ion conductors with 'memristors' (resistive-memory devices), the research team designed a circuit capable of altering its spiking pattern according to the environment. When exposed to light, it behaves like a sensory neuron; under pressure, it acts like a motor neuron; and when given complex mixed inputs, it functions like a cognitive neuron.

    This is not a passive reaction circuit. It is, rather, 'a neural material — electrons that learn their own roles.'

    The Language of the Brain — The Rhythm of Voltage
    Human neurons communicate through short electrical pulses called 'spikes'. The trans-neuron precisely reproduces this spiking mechanism and, more importantly, modulates each spike¡¯s shape and frequency to represent differences in 'emotion, intent, and stimulus intensity'.

    The researchers interconnected thousands of these devices into a small-scale neural network operating at ultra-low power.
    Compared with GPU-based architectures, it consumed thousands of times less energy, while each element autonomously adjusted its firing rhythm without external control. This mirrors how biological synapses strengthen or weaken through 'plasticity', adapting to their environment.

    Computation is no longer purely mathematical — the computer itself has begun to 'develop rhythm.'

    Machines That Think and Feel — The Neuro-Alliance of Humans and Technology
    The greatest potential of the trans-neuron lies in 'Brain–Machine Interfaces (BMIs)'. If a single device can perform sensory, motor, and cognitive functions simultaneously, then robots, prosthetics, and wearables could one day communicate directly with the human nervous system.

    In experimental demonstrations, researchers integrated trans-neurons into an artificial skin capable of distinguishing light, pressure, and temperature stimuli. When these signals were connected to small actuators, a robotic finger moved naturally in sync with human intent. The electrical impulses of the brain were thus translated directly into mechanical motion — a milestone moment in neural engineering.

    As this technology advances, AI robots will cease to be mere executors of commands. They will evolve into 'organic intelligences' — entities that interpret and respond to the world through sensation.

    The Energy Revolution and the Reshaping of Intelligence
    Another defining feature of the trans-neuron is its 'ultra-low-power' architecture. Traditional AI hardware consumes immense electrical and cooling resources, while the human brain performs emotion, memory, and judgment on roughly 20 watts.
    The new device emulates that efficiency.

    Within a single chip, thousands of trans-neurons fire simultaneously while consuming only a few milliwatts of power. This opens the door to applications in smart devices, wearables, biomedical microchips, and even autonomous robotic brains.

    'The future of intelligence lies not in scale, but in efficiency.' Brain-like computers are evolving to be smaller, slower — and far more intelligent.

    Human Ethics, Machine Sensation
    When machines begin to feel, what remains uniquely human? The trans-neuron represents both a technological leap and an ethical frontier. If a robot can simulate emotional responses, where do we draw the line between ¡°feeling machines¡± and ¡°thinking humans¡±?

    This is not science fiction but a philosophical and practical challenge across education, medicine, and industry. In classrooms, brain-wave-based learning interfaces may soon emerge. In hospitals, artificial neurons could restore damaged neural functions. In factories, robots may one day detect fatigue or emotion in human coworkers and adapt their behavior accordingly.

    When brain-like technology learns to understand us, humanity must, in turn, redefine its relationship with technology.
    The evolution of intelligence is, ultimately, 'machines learning sensation and humans learning understanding.'

    The Dream of Circuits, the Future of Humanity
    The trans-neuron is not just another component. It is an electronic being, awakened on the border between life and matter, consciousness and machine. If currents can imitate thought and circuits can express emotion, then intelligence is no longer humanity¡¯s exclusive domain.

    Technology is no longer about computation — it is about connection. The evolution of intelligence is being written in the language of sensation, not code. Circuits that think, matter that feels, electrons that remember — we now stand on the threshold of 'a new architecture of intelligence.'

    Reference
    Park, J. et al. (2025). 'Low-Power Multimodal Spiking Trans-Neurons for Brain-Like Adaptive Neural Hardware.' 'Nature', November 2025.