Smell was long considered the only sense that had escaped digitalization. Smell-O-Vision remained an aspiration rather than a reality, with a plethora of mysteries remaining around how to digitally interpret the complex human olfactory system. However, a new influx of applications and scientific discoveries are looking to drive the digital olfaction industry segment forward—a market that is set to reach a worth of $1bn by 2025.
The backlog of innovations thus far presents a promising future. In 2014, the oPhone introduced a system of scent-based mobile messaging, whereby users could send olfactory impressions to one another via an app and mobile dock system which emitted selected odors from a database of 10,000 different olfactory impressions. Dr. Jenny Tillotson’s eScent invention fused accessories with timed scent diffusion, while IBM’s Philyra AI algorithm simplified the scent creation process. The most recent invention has been the NeOse Pro, which fuses surface plasmon resonance imaging technology for odor distinction with 100 biosensors in order to detect and identify 500+ odors. The product, which retails for €10,000, has been purchased by leading fragrance companies worldwide, but can also be employed in other tasks such as food quality control, pollution monitoring, and auto vehicle maintenance. With a recent securement of €7MM in funding, the product will be refined and adapted to allow for greater distribution and application amongst a variety of industries.
One of the main hurdles standing in the way of the digital olfaction market had been the refinement of scientific underpinnings to fuel a computerized olfactive system. Recent discoveries by esteemed research labs may offer the insight to overcome this. An experiment by the Rinberg Lab at NYU found that phantom smells could be created in mice brains through the use of light, eschewing any interaction with the olfactory centers of the brain. While of course, human brains are much more complex, this discovery could potentially be harnessed in the favor of recreating scents in the absence of the smell itself.
Meanwhile, researchers at Google Brain are teaching machine-learning algorithms to determine molecules’ olfactory character via their chemical makeup by employing a graph neural network. The machines were given close to 5,000 molecular structures to work with, deciphering the minute differences between each one. However, there are still discrepancies between how the machine and how a human interprets scent, such as mirrored molecules, which the machine reads as the same despite the fact that they smell vastly different (the example cited by the Google team was spearmint and caraway).
The complex neural networks underlying olfaction also present a new chapter for coding artificial intelligence, argues Jordana Cepelewicz on Wired.com. An AI algorithm devised by Cornell University and Intel, complete with 130,000 digital “neurons,” can accurately identify 10 different smells. “It’s the basis for a set of devices for artificial olfactory systems that can be constructed commercially,” Thomas Cleland, professor of psychology at Cornell, states about the resulting system.
A team of scientists at the University of California, Riverside, have created an AI that can predict how a chemical will smell to humans, overhauling how these materials can be used in the food, fragrance, and flavor industries. For example, a chemical in a product that would be perceived as a harsh smell by consumers could be identified and replaced with a more delicate-smelling one at a much earlier stage in the product development process than before.
Outside of the scientific realm, digital olfaction would provide a huge boost to fragrance e-commerce. While perfumery was still heavily reliant on in-store experiences, the COVID-19 pandemic forced the industry to rethink its strategies. Samples have previously been the method of at-home experimentation for consumers prior to purchasing a full-size product, but “sending” these in digital form would significantly reduce carbon emissions. Unfortunately, digital smells thus far have been limited to singular or more basic constructions, rather than the complex creations that are consumer fragrances. “I believe digital is an opportunity which brands—and retailers—cannot afford to miss out on today,” CEO of travel retail consultancy TW.O & Partners, Fabio Bernardini, tells Cosmetic Business. AI consultations and AR try-on simulators have already been employed in the online beauty markets, and while the former could seamlessly translate into the world of olfaction, the latter would be heavily reliant on the evolution of digital olfaction technology. It is difficult to determine exactly how and when these technologies will be fit for at-home consumer consumption, but judging by recent developments, the era of digital olfaction advancing a variety of industries is imminent.
