Jakarta, INTI - The field of artificial intelligence is entering a new era. This time, scientists are no longer relying solely on algorithms and silicon chips, but are incorporating living neurons into computational systems. This approach has given rise to a new concept known as bio-AI, a fusion of biological neural networks and machine learning technology.
In a recent study, researchers successfully developed a system that uses neurons from the rat cortex to perform computational tasks directly. This discovery opens up the possibility that biological neural networks are not only components of living organisms, but can also function as information-processing “machines” similar to computers.
Integrating Living Neurons and Machine Learning
The core idea behind bio-AI is the integration of living neurons with machine learning systems within a single ecosystem. The neurons used are extracted from the rat cortex and then “trained” to respond to specific signals and produce measurable outputs.
Unlike conventional AI, which is entirely software-based, this system operates by leveraging the natural electrical activity of neurons. This means computation is no longer dependent solely on code, but also on complex biological dynamics.
This approach offers a distinct advantage. Living neurons have a high degree of adaptability, allowing them to learn and adjust in real time without requiring manual reprogramming.
How It Works: Closed Loop Reservoir Computing
The technology used in this research is based on the concept of reservoir computing. In this method, the neural network acts as a “reservoir” that transforms input signals into complex patterns.
These neurons are connected to a microelectrode array and a microfluidic system. This setup records neuronal electrical activity, converts it into digital data, and then feeds electrical stimulation back into the system.
This creates a closed-loop system, where the output of the neurons becomes the input for the system itself. The cycle operates extremely quickly, with a delay of around 330 milliseconds, enabling continuous learning.
Remarkably, the entire process runs automatically without human intervention. The system can adjust its responses based on incoming patterns, similar to how the brain learns from experience.
Micro Scale Network Design for Efficiency
To improve performance, researchers did not rely on randomly arranged neurons. Instead, they designed a structured micro-network consisting of 128 small compartments connected by microchannels. This design helps reduce a common issue in biological networks, excessive synchronization.
Under normal conditions, neurons tend to fire simultaneously, which reduces information complexity. By separating them into smaller compartments, neural interactions become more diverse and dynamic. As a result, the system produces more complex and efficient signal patterns.
Ability to Generate Complex Patterns
One of the key achievements of this bio-AI system is its ability to generate various waveform patterns. These range from simple signals such as sine, square, and triangle waves to more complex structures like the Lorenz attractor.
This capability demonstrates that biological neural networks can be used to model complex dynamic systems. In tests, the system achieved relatively high accuracy, with correlation values exceeding 0.8. This indicates strong potential for applications requiring high flexibility and adaptability in computation.
Remaining Challenges
Despite its promise, the technology still faces several limitations. One major challenge is performance degradation after the training phase stops. Without continuous stimulation, the neurons’ ability to maintain learned patterns tends to weaken.
Another issue is the feedback delay of approximately 330 milliseconds, which limits its use in applications requiring extremely fast response times. These challenges are the focus of ongoing research aimed at improving system stability and speed.
The Future of Bio-AI
Looking ahead, researchers plan to develop specialized hardware that reduces latency while improving neural communication efficiency. If successful, bio-AI could become the foundation of a new generation of artificial intelligence systems.
Its potential applications are broad, ranging from brain–machine interfaces and neural prosthetics to hybrid AI systems that combine biological and digital intelligence. With these advancements, the boundary between machines and living organisms is becoming increasingly blurred. Bio-AI is not just a technological innovation, but a step toward a new paradigm in computing.
Conclusion
Bio-AI represents a groundbreaking shift in the field of artificial intelligence by integrating living neurons with machine learning systems. This hybrid approach enables real-time learning, high adaptability, and complex signal processing beyond conventional AI models. Although challenges such as system stability and latency remain, ongoing research shows strong potential for future applications, including brain machine interfaces, neural prosthetics, and next-generation intelligent computing systems. Ultimately, bio-AI signals a new era where the boundary between biological systems and digital machines continues to blur.
Read more: Japan Wants to Build a Solar Ring Around the Moon to Generate Energy for Earth