The researchers at Google form the Google Brain Team have created a dataset of nearly 5,000 molecules identified by perfumers and then used about 66% of it to train their Artificial intelligence to associate molecules with related descriptors. The scientists then used the remaining 34% of the scents to test the AI and it was accuractely able to identify molecules smell based on their structure.
Also Read: Escape This Haunted House And Win $20,000 In Prize Money
Google further stated that molecule descriptions ranged from “buttery” to “tropical” and “weedy”.
Read more at Engadget