Gin created by artificial intelligence

The gin Monker’s Garkel uses juniper berry, coriander seed, angelica root, raspberry leaf, gooseberry, prune, clementine zest, orange zest, and marigold: and has an ABV of 40%.

A collaboration between technology scientists Tiny Giant and Rewrite Digital created Ginette: a neural network that is the ‘brains’ behind Monker’s Garkel.

Trained to analyse thousands of botanicals, understand gin recipes and learn a database of more than 500 gin names, Ginette went through a process of creating a huge catalogue of brand names, designs and botanical recipes before deciding on ‘Monker’s Garkel’.

Ginette also had a say in the label and marketing. An automated algorithm created the labels using a curated set of styles, colours, textures and fonts, which was then formulated by local artist Tim Sutcliffe.

The copy on the label and Ginette's communication on social, @GGinmaker is written by a neural network based on the Open AI GPT-2 language model.

Liam Hirt, co-founder of Circumstance Distillery, said: “AI and machine learning are exciting new tools. We do a lot of contract distilling and wanted to know if these tools could make a meaningful contribution to new product development. ​

"At present they are not a substitute for the skill of the distiller, but we have clearly shown that they can make a valuable contribution to all aspects of the development process, from the recipe to the naming and labelling. ​

"We trained Ginette well, and with a little human help and guidance, her product is interesting, attractive and, most importantly, delicious.”​

Monker's Garkel was launched at the Bristol Technology Festival earlier this month, and will be served at MAD//FEST in London next week. 

Earlier this year, Swedish distillery Mackmyra launched the world's first whisky created using artificial intelligence in a collaboration with Finnish technology company Fourkind, based on the Microsoft Azure platform.