In the best-case scenario, two identically named craft beers just means creators feeling despondent that their whimsically titled “Raven’s Goblin” ale is not quite as unique as they thought. In the worst case, two identical names could mean marketplace confusion and, potentially, lawsuits.
Shane said she started out her project by getting help scraping more than 100,000 beer names, representing more than 90 subcategories, from the website BeerAdvocate.com. She then sorted these into major subcategories and began the neural net training process. “I trained the neural network separately on IPAs, strong ales, red ales, and stouts, ending up with four specialized models,” she told Digital Trends.
The names the neural net has so far created range from the inspired (an amber ale called Blangelfest) to the odd (a stout called Spulgican’s Chocolate Coconut Pamper) to … ones that sound like they are describing something else entirely (O’Busty Irish Red). Shane’s favorites? Three red ales called Snot Beard, Slugsty Red Ale, and Wooly Barrel Choo Red.
“I really like the way applying AIs to creative tasks gives us a look at ourselves from a brand-new perspective,” she said. “What does a computer make of our world? What patterns does it discover that we had no idea existed, and what patterns does it miss? We rely to such an extent on our background and cultural knowledge when dealing with the world that it’s fun to see what happens when this is stripped away.”
And, hey, doesn’t a good craft beer make everything a bit more fun? Unless it’s called Snot Beard, that is.
Editors' Recommendations
- Revisiting the rise of A.I.: How far has artificial intelligence come since 2010?
- Machine learning? Neural networks? Here’s your guide to the many flavors of A.I.
- With names like Burf Pink and Bank Butt, a paint algorithm fails spectacularly
