Russian Neural Network Is Being Trained to Distinguish Smells

Researchers at the Far Eastern Federal University are training a neural network to identify odors, the university’s press service said. Scientists at the Institute of High Technologies and Advanced Materials explained that today’s artificial intelligence is largely a set of algorithms designed for specific tasks. In the future, however, a higher-level form of intelligence — artificial general intelligence (AGI) — could emerge, capable of processing large volumes of data, understanding context, learning independently, and even exhibiting analogues of human emotions.

“Breathing” Through Laser Sensors

According to project leader Vladimir Lisitsa, the “digital nose” will perceive information using laser-based sensors. This technology can detect substances in any physical state and identify elements across the entire periodic table. As a result, the AI’s sense of smell would be “incomparably richer than that of humans,” he said.

The FEFU researchers aim to combine several neural networks into a kind of olfactory center. This could effectively give artificial intelligence a form of “sensory perception.” The technology could be applied across a wide range of fields, from medicine to industry. Smell-based analysis could be used to diagnose diseases, detect specific substances in the air, and even create new fragrances.