top of page
  • Writer's pictureLuke Nyswonger

Designing A Grimace-Inspired Purple Milkshake IPA with ChatGPT

There's something magical happening at McDonald's. The fast-food giant has reintroduced Grimace, the endearing purple mascot who has been absent from our screens for years. Initially known as the 'evil milkshake-stealing monster,' Grimace has made a jovial comeback, and he's not empty-handed! McDonald's is marking Grimace's birthday with a brand-new creation - the Grimace Shake, a captivating purple milkshake with a unique blend of flavors reminiscent of 'purple Froot Loop' and 'blueberry' entwined with vanilla. Being a kid of the 70s and 80s, a time when both Chicken McNuggets and Grimace were introduced to me, I can truly say 'I'm lovin' it'!


Grimace with a shake sitting on a car
Official image of Grimace, the inspiration behind our latest brew. Image credit: McDonald's.

The link between the iconic Grimace, his namesake milkshake, and homebrewing may seem tenuous at first, but allow me to shed some light. The spirit of homebrewing is inherently linked to creativity, innovation, and exploration. It's about translating our wildest imaginations into tangible, palatable experiences. The latest McDonald's campaign, showcasing the quirky Grimace and his intriguing purple shake, serves as a muse for this creativity, fueling our innovative journey in brewing.


I'm not just replicating the Grimace Shake; I'm reimagining it in the form of a Milkshake IPA, another beloved concoction that marries sweetness, creaminess, and hoppiness in a delightful symphony. As an AI enthusiast and a passionate homebrewer, this Grimace-inspired campaign has provided an exciting challenge - create a version of this using my existing homebrewing knowledge and ChatGPT. The aim is not just to create a replica but to encapsulate the spirit and flavor profile of the Grimace Shake in a beer format. I'm infusing this adventure with vibrant purple hues to celebrate Grimace and nod to the festivities of Pride Month in June.

Grimace's Pride Milkshake IPA
Bing Image Creator: Grimace's Pride Milkshake IPA

Understanding Milkshake IPAs


Before we proceed, let's get a grasp on what a Milkshake IPA is. This style of beer is a subset of the New England IPA, recognized for its hazy appearance and fruity hop character. What sets Milkshake IPAs apart is the addition of lactose (milk sugar) which not only adds a creamy sweetness to the brew but also helps create a fuller, smoother mouthfeel. The flavor profile is typically rich, with pronounced fruit notes and a soft, balanced bitterness. It's not unusual to find actual fruits added in the brewing process, amplifying the juicy, fruity complexity.


Milkshake IPAs are a far cry from your ordinary Coors Light; they boast a rich, creamy thickness that can make even the sturdiest of pint glasses feel as though they're cradling a dessert rather than a brew.


Designing the Grimace-Inspired Milkshake IPA


Now onto the fun part: designing our Grimace-inspired Milkshake IPA. We're going to use the ChatGPT-4 model connected to Bing to help us figure out the best way to construct this recipe, keeping in mind that we want to create a beverage that mirrors the look and taste of the Grimace Shake and comes in at a robust 8%+ ABV.


Starting with the base of our IPA, I'll opt for a simple 2-row that lends a light and pleasant base for the other ingredients to shine. While I won't specifically tell the AI model what base grains I want, I'll be descriptive enough that it should pick them out, including the use of flaked Oats. In line with the Milkshake IPA style, I'll tell ChatGPT to add lactose during the brewing process to get that creamy mouthfeel.


For the all-important fruity character, we have some options that I could see but I think using blackberries might be an interesting twist to this and so I'm going to constrain the model to use that fruit. Not only will these delicious berries provide a wonderful sweet-tart flavor that melds beautifully with the IPA's hop character, but they'll also lend our brew a stunning purple hue.


Let's prompt ChatGPT to nail down our recipe. In my article "Prompt Engineering 101 for Homebrewers", I talk about exploratory versus directive prompting. The prompt I'm using for this one is a blend of both:


Prompt: Help me design a purple-colored Milkshake IPA recipe. I'll be adding lactose for creaminess and blackberries for flavor and color. Make sure to incorporate vanilla as well. I'm aiming for an ABV of 8% and brewing a 5-gallon batch using all-grain.

From an exploratory perspective, the prompt allows the AI model to consider a wide range of possible responses as it designs a Milkshake IPA recipe. The AI is allowed the liberty to explore the best way to construct the recipe considering the ingredients and the desired attributes of the final product.


On the directive side, the prompt provides specific guidelines or constraints for the AI to adhere to. It directs the AI to ensure that the recipe is purple-colored, uses lactose for creaminess, blackberries for flavor and color, incorporates vanilla, and results in an ABV of 8% for a 5-gallon batch using all-grain. These directives limit the scope of the AI's responses, ensuring the output aligns with the specific brewing goal.


A significant number of criticisms surrounding the performance of generative AI can, interestingly enough, be attributed not to the AI itself but to the quality of the input it receives. The effectiveness and accuracy of an AI's output are, to a large extent, determined by the precision and clarity of the user's prompts. In essence, a generative AI is a tool, and like any tool, its efficacy relies heavily on how it is used. A well-defined and thoughtful prompt can lead to valuable and creative results, while a vague or overly simplistic input may result in an output that's less than satisfactory. Something like this:

Prompt: Create a beer recipe

The first prompt, which can certainly be improved upon, is a good blend of exploratory and directive prompting which leverages the strengths of both approaches, providing enough flexibility for creative solutions while maintaining the necessary constraints for a specific brewing objective. It's an excellent illustration of how nuanced and carefully crafted prompts can guide the AI to generate relevant, creative, and useful outputs.


ChatGPT-Crafted Recipe


So, what was the output?


In this experiment, I'm embracing the spontaneity and creativity of AI by committing to the first output generated by ChatGPT. No iterations or multiple rounds of prompting; I'm capturing the raw, initial output. However, as any experienced brewer knows, brewing is an art form, and often adjustments are made in the heat of the moment. While our base recipe remains unchanged, I may tweak things during the actual brewing process.


The initial output from ChatGPT wasn't terrible. You can see below that the base of Pale 2-Row malt, Flaked Oats, Wheat Malt, and Honey Malt, along with Lactose, should provide a creamy, full-bodied mouthfeel and a sweet backbone that emulates the dairy base of a milkshake. I'm still debating whether or not I'll use the Honey malt.


Citra, Mosaic, and Galaxy hops should introduce a vibrant, tropical, and berry-like hop character that complements the fruity addition. And the 5 pounds of blackberries used in secondary fermentation not only imbue the beer with a purple hue reminiscent of Grimace but also infuse it with a natural berry flavor that mirrors the taste described in the Grimace shake. One thing I will change is the yeast. I'm a fan of Imperial Yeast and will either do Barbarian or Juice.

 

Presenting to you, the unaltered recipe directly from ChatGPT. Apart from the adjustments I previously noted, I plan to brew this concoction exactly as it was generated.


💜Grimace's Pride Milkshake IPA


Batch Size: 5 gallons


Ingredients:

  • 12 lbs Pale 2-Row Malt

  • 1 lb Flaked Oats

  • 1 lb Lactose

  • 1 lb Wheat Malt

  • 1 lb Honey Malt

  • 1 oz Citra Hops (60 minutes)

  • 1 oz Mosaic Hops (15 minutes)

  • 1 oz Galaxy Hops (0 minutes)

  • 2 oz Mosaic Hops (Dry Hop at 3 days)

  • 2 oz Galaxy Hops (Dry Hop at 3 days)

  • 5 lbs Blackberries (secondary fermentation)

  • 2 Vanilla Beans, split and scraped (Secondary Fermentation)

  • 1 pkg London Ale Yeast (such as Wyeast Labs #1028)

Procedure:

  1. Mash the grains at 152°F (66°C) for 60 minutes.

  2. Proceed with a 60-minute boil, adding Citra Hops at the start of the boil.

  3. With 15 minutes left in the boil, add the Lactose and the Mosaic Hops.

  4. At the end of the boil, add the Galaxy Hops.

  5. Chill the wort quickly to around 70°F (21°C), then pitch the London Ale Yeast.

  6. Ferment at the recommended temperature of the yeast.

  7. On the third day of fermentation, add the first round of dry hops (Mosaic and Galaxy).

  8. After the primary fermentation is complete (usually about one week), add the blackberries to the secondary fermenter and rack the beer onto the fruit.

  9. Allow the beer to sit on the fruit for another week, then add the second round of dry hops.

  10. After another 3-4 days, check the gravity to ensure fermentation is complete, then you're ready to bottle or keg!


Conclusion

And there you have it - a unique purple Milkshake IPA, co-developed with AI and inspired by a beloved fast-food mascot and the spirit of Pride Month. I hope you've found this journey as fascinating as I have. At Homebrewing.ai, I explore the intersections of AI, homebrewing, and technology. If you enjoyed this glimpse into the possibilities that AI offers homebrewers, do consider subscribing to the blog for more insights and experiments. For lively discussions, sharing experiences, and to stay on top of updates, join our growing community on our Facebook page.


Grimace might be a fan of milkshakes, but we suspect he'd trade his ice cream for a pint of this brew any day! Cheers! 🍻

1,654 views3 comments
bottom of page