When you see an image on your computer screen, it is technically a combination of only three ingredients: green, red and blue.
But what if the computer could recreate an odour? It is a much more complicated matter, and there is still much we do not know about odours.
But now, researchers from Google Research at Cambridge University have used artificial intelligence (AI) to create an odour map based on the odours’ molecular structure, according to New Scientist. The research is supported by Mark Zuckerberg through the Chan Zuckerberg Initiative.
The technology is thus one step closer to being able to identify what makes strawberries smell like strawberries and coffee like coffee. A preprint study shows that the algorithm is about as good at identifying odours as humans, and in another study, the researchers show how it can be used for useful purposes such as testing new types of mosquito repellent.
Odours are far more complicated than colours and sounds because they do not appear to consist of any “primary” components.
Our noses have over 300 different olfactory receptors. The slightest difference between the structure of two molecules can make a world of difference in their smell, and it is a combination of many molecules that makes strawberries smell like strawberries.
The researchers created the map by training the AI model on 5,000 different molecules whose odours and flavours are described in two man-made databases with words like “grassy” or “creamy”. After this, the AI had to try to predict the smell of 400 new molecules. It hit the spot on 320 of them, about as many as the 15 people who participated in the study.
The study is a significant advancement in the research field that investigates smell, believes Professor Barry Smith from the University of London, who talked to New Scientist. He emphasizes that there is still a long way to go to understand how smell works.
“What the work does is skip neurobiology and try to connect the structure of molecules to the perception of odours directly. Impressive if it can be done, but we will still have to fill in the biology eventually if we want to understand how humans perceive odours.”
The researchers hope that the technology will become so advanced that it will in the future be possible to create new scents from scratch based on the new knowledge about the molecules.
That would not only be useful for the perfume industry, but it could also be used on in a broader sense, the researchers believe.
In the second study, the researchers investigated how artificial intelligence can help create new types of mosquito repellent that make more mosquitoes stay away.
This suggests that the technology will also be able to help develop substances that can scare other insects away.
Since insect-borne diseases are a major threat to human life in large parts of the world, the research may one day help save lives, the researchers behind the study hope.
Vi bygger bro med stærke vidensmedier, relevante events, nærværende netværk og Teknologiens Jobfinder, hvor vi forbinder kandidater og virksomheder.
Læs her om vores forskellige abonnementstyper
Med vores nyhedsbreve får du et fagligt overblik og adgang til levende debat mellem fagfolk.
Teknologiens Mediehus tilbyder en bred vifte af muligheder for annoncering over for ingeniører og it-professionelle.
Tech Relations leverer effektiv formidling af dit budskab til ingeniører og it-professionelle.
Danmarks største jobplatform for ingeniører, it-professionelle og tekniske specialister.
Kalvebod Brygge 33. 1560 København V
Christina Blaagaard Collignon
Trine Reitz Bjerregaard