IBM Hypertaste e-tongue uses AI to rapidly identify liquids

IBM has revealed Hypertaste, an 'e-tongue' that can rapidly pick out the flavors in liquids. The company describes its creation as an AI-powered electronic tongue, one that can be used to fingerprint liquids that are less than suitable for ingestion, but that need to be identified more rapidly than is possible using traditional laboratories. Examples include remote water quality checks and testing for counterfeit alcohol.

Though it's possible to analyze liquids by sending them to a lab, this process requires quite a bit of time and money. IBM Hypertaste is presented as an alternative to this approach, making it possible to see the composition of a liquid in a single minute on a smartphone.

The e-tongue utilizes something called combinatorial sensing, which involves individual sensors simultaneously responding to different chemicals in a liquid. The result is a 'fingerprint' of the liquid. The device is made using off-the-shelf electronics and works with a mobile app to transfer data to the cloud.

Once the data is on the server, a machine learning algorithm provides results on the user's phone based on a database of other liquids. Different machine learning models can be used, according to IBM, making it possible to use the same e-tongue device for different tasks.

Hypertaste could prove useful in a variety of industries, including the food and beverage industry, environmental work that takes place in remote areas, and the healthcare and pharmaceutical industries. Packaged beverages could be rapidly tested on-site to determine whether they actually contain the beverage shown on the packaging, for example, something made possible by the rapid, portable nature of the e-tongue and mobile app system.