Osmo is an AI research company that built the Principal Odor Map, a machine learning model trained to predict how any molecule will smell based solely on its chemical structure. Their research, published in Science in 2023, demonstrated AI can predict smell with accuracy comparable to trained human experts. The long-term goal is a digital representation of the olfactory world: encoding smell the way MP3s encode sound. Founded by former Google Brain researcher Alex Wiltschko, Osmo is working on practical applications including AI-designed fragrances that avoid harmful chemicals, faster drug development, and improved air quality monitoring systems. The Principal Odor Map is the first step toward a searchable, shareable standard for olfactory experience. The scientific AI community engaged heavily with Osmo's Science paper on Hacker News and in academic discussion threads, calling it a landmark result in a neglected sensory domain. The broader public finds 'AI smells' equal parts fascinating and baffling. Still pre-consumer-product for most users, but the implications of digitizing the most underexplored human sense have generated genuine excitement. Based on community discussions from Hacker News and scientific Twitter.

What the community says

The Science paper generated one of the more enthusiastic Hacker News threads of 2023, with researchers calling it a genuine milestone in a scientifically neglected domain. The concept of 'digitizing smell' captures strong public imagination. Skeptics note the lack of a consumer product and question near-term timelines. The science itself is not in dispute; the commercial roadmap is. Based on community discussions from Hacker News and scientific discussion forums.