Can Google smell? Why the digitization of smell could be a business opportunity

Artificial intelligence is entering the scent business. Google researchers recently announced that they trained an AI-powered neural network model to “map” how a molecule The structure correlates with its aroma and could cause a revolution in the fields of food, fragrance and health.

Say, for example, your company makes bug spray. The Google researchers found that by feeding their neural network data about how effective various molecules are at repelling mosquitoes, the resulting model can go on to predict the mosquito repellency of almost any molecule. Humans are able to smell things because the tiny molecules are processed by receptors in your nose, which then send a message to your brain. The researchers found that more than a dozen tested molecules showed repellency at least as strong as DEET, the active ingredient in most insect repellents. These molecules could form the basis for less expensive, longer-lasting and safer spraying.

Google researchers say smell is the most difficult of the senses to quantify in machine-understandable data, and they’ve been working for several years to train their AI models to predict how molecules will smell by analyzing their structural makeup. By feeding the network with data containing the composition of over 5000 molecules, combined with multiple aroma descriptors for each molecule, the researchers were able to create what they call an “odor master map”. The map identifies the relationships of different odors to each other by graphically forming thousands of data points, each representing the structure of a single molecule analyzed by the neural network.

In practice, the map works like a color wheel, with similarly structured molecules grouped close together in the same way that similar colors such as red and orange would be grouped together. The researchers found that nearby molecules tended to have the same scent descriptions.

In a test where trained participants were asked to identify the odors of 400 different molecules using 55 descriptive labels, the neural network was able to accurately predict the panel’s consensus responses far better than any individual participant. The neural network was also able to measure the strength of specific odors.

The researchers say the AI-generated map enables them to predict how billions of unknown molecules will smell, which could have “broad applications in flavor and aroma”.

Leave a Comment