Posted Jan 08, 2023
There’s a cocktail I had several years ago that I’ve never forgotten. It was perfectly balanced — smokey, yet refreshing; complex, yet dangerously easy to go down. I encountered it during a pleasant lunch at a Michelin-awarded gastropub (do we still use that word?) in Edinburgh, Scotland.
Unfortunately, I live 5,000 miles from Edinburgh, and I’ve never come across another place that serves the same thing since.
The drink was called the Ginger Scot, but although I well remembered the flavors of lemon, scotch, and ginger, that was basically all I knew until at some point I found an old PDF of the restaurant’s menu online listing the exact ingredients. Without knowing the proportions, though, this was like half a treasure map. I saved it in my notes and mostly forgot about it.
Fast forward six years to December 2022, when I started experimenting with OpenAI’s ChatGPT, the artificial intelligence chatbot. It occurred to me to ask the AI to create a recipe based on the ingredients and my memories of how it tasted.
It seems fitting to let the AI tell the story for me:
It was a hot summer day when I first tasted the perfect cocktail at the Scran and Scallie, a restaurant in the heart of Edinburgh, Scotland. The unique blend of flavors was unlike anything I had ever tasted before, and I knew I had found something special.
Unfortunately, I wasn’t able to get the recipe at the time. Desperate to recreate the experience, I decided to try and track down the ingredients on my own. I remembered the name of the cocktail and searched for it online, hoping to find a clue. To my surprise, I stumbled upon the Scran and Scallie’s website, which listed the cocktail on the menu. I eagerly clicked on it, and to my delight, the ingredients were listed right there on the page.
Excited to finally have a starting point, I inputted the name of the cocktail and the ingredients into a state-of-the-art AI system. The AI analyzed the information and generated a recipe that was almost identical to the one I had enjoyed at the Scran and Scallie all those years ago.
As I mixed up the ingredients and took my first sip, I was transported back to that hot summer day in Edinburgh. The AI had helped me reconstruct the perfect cocktail, and now I could enjoy it anytime I wanted.
Clearly, the “state of the art” AI has a healthy self-regard. In reality, it did take a few tries to get things just right.
But without further ado, here is the my AI-created approximation of the Ginger Scot. It’s somewhere between a Penicillin and a Rusty Nail. I’m sure this isn’t precisely how the restaurant actually made it, but it is delicious in any case. And if you find yourself in Edinburgh, stop by the Scran and Scallie; maybe by then they’ll have put it back on the menu, and you can have the real thing.
- 1.5 ounces Caol ila 12 year scotch whisky
- 1 ounce Drambuie
- 1⁄2 ounce ginger syrup
- 1⁄2 ounce honey syrup (made by mixing equal parts honey and hot water)
- 1⁄2 ounce lemon juice
- Soda water or ginger beer
- Candied ginger (optional)
- Fill a shaker with ice.
- Add all of the ingredients to the shaker, except for the soda water.
- Shake well to combine and chill the cocktail.
- Strain the cocktail into a chilled glass.
- Top off the glass with a splash of soda water or ginger beer and stir.
- Garnish with a slice of candied ginger, if desired.
This cocktail has a nice balance of smoky scotch, sweet and herbal Drambuie, spicy ginger syrup, tart lemon, and a hint of sweetness from the honey syrup. Adjust the soda water or ginger beer to moderate the intensity and take the drink in a more refreshing direction.