Soon, your computer will be able to smell you
Artificial intelligences are developing at a dizzying speed, but if there is one thing we are clear about is that they lack life. In every sense—well, in some “senses” more than others. With the advancement of this technology, AI models are now capable of seeing and listening, but will they be able to “feel” something more?
One of the main objectives of Osmo is to use its technology to help medical professionals in disease detection, leveraging the information contained in odors. However, in the short term, the company focuses on creating safer and more sustainable aromatic molecules for everyday products such as perfumes, shampoos, insect repellents, and laundry detergents. According to Osmo, the current fragrance market is dominated by a few companies, and their technology could revolutionize the sector by designing ingredients that are non-toxic and non-irritating to skin or eyes.
The innovation of Osmo lies in the use of machine learning and graph neural networks (GNN) to create a “main odor map.” They trained their AI model with data from 5,000 aromatic molecules, successfully predicting the smells of new molecules. However, this process was quite a challenge, as molecular structures are complex, and even the smallest variations can significantly alter the smell.
Osmo, a startup founded by Alex Wiltschko, aims to digitize the sense of smell using AI, with the ambitious mission of improving human health and happiness. As reported by CNBC, the company is developing a new type of technology that would allow computers to generate and detect smells, similar to how sight and hearing are already imitated in AI systems.
Looking ahead, Osmo intends to develop a new type of technology capable of teleporting smells, digitizing an aroma in one place and recreating it in another. In addition, the company continues to work towards its long-term goal of using digital smell to detect diseases, a development that, although still a long way off, promises to revolutionize medicine.
