A Computer Can Taste Wine?

In the vast world of wine enjoyment, tasting notes are in some ways akin to a compass, guiding wine lovers through a never ending list of possible flavors, aromas, and sensations. They capture ephemeral moments of pleasure, translating them into words that can be shared and savored. Today, technology is offering a new twist. With the advent of Artificial Intelligence and LLMs, it's now possible to transform our tasting experiences into detailed notes and reviews, even if you're not a wordsmith or a seasoned sommelier. In this post, we'll explore how an AI author can take your sensory inputs and weave them into an informative and evocative wine review.

It's important to mention that the richness and precision of the AI-generated tasting note depend heavily on the quality and detail of the human input. The more specific and descriptive you can be about the wine's characteristics, the better the AI can generate a comprehensive review.

For instance, consider a human input to ChatGPT about a red wine: "The wine is dark red, almost purple. It smells like cherries and a hint of vanilla. It tastes dry and has a long finish." This brief description gives a good start, but it's still quite basic.

From this input, I, as an AI author, could generate the following tasting note: "The wine exhibits a deep red, verging on purple, color, indicative of its concentrated intensity. On the nose, the wine reveals charming aromas of ripe cherries, accompanied by an underlying whiff of vanilla, suggestive of oak aging. On the palate, it presents itself as a dry wine, culminating in a lengthy finish that lingers, affirming its quality."

However, to elevate the quality of the tasting note, we could enhance the human input with more details about the wine's characteristics. For example, commenting on its body, acidity, tannin level, and the intensity and complexity of the aromas and flavors could yield a much more informative tasting note.

Let's try a more elaborate human input: "The wine is dark red, almost purple. It has a strong smell of ripe black cherries, blackberries, and there's a hint of vanilla and spice - maybe clove? It's dry, with high acidity and medium tannins. The fruit flavors are very bold, and there's a sort of peppery taste too. It's full-bodied, with a long finish."

With this richer input, an AI-authored tasting note might read: "This wine commands attention with its deep, almost purple hue, suggestive of its potent intensity. The nose is striking, overflowing with the lush aromas of ripe black cherries and blackberries. An inviting undertone of vanilla weaves in, the likely result of careful oak aging. A subtle spice note hints at the presence of clove, adding complexity to the aromatic profile.

On the palate, this wine doesn't disappoint. It's dry, with an invigorating high acidity that beautifully complements its medium tannin structure. Bold fruit flavors mirror the nose, dominating the palate with a harmonious echo of black cherries and blackberries. A dash of pepperiness adds an intriguing twist, ensuring this is a wine that continues to surprise and engage.

The body of this wine is unapologetically full, filling the mouth with its rich and robust character. It concludes with a long finish, a final act that leaves a lasting memory of its superior quality."

An AI author is fully capable of crafting a wine review or tasting note based on the inputs from the person who tasted the wine. The more descriptive and detailed the human inputs, the richer and more accurate the AI-generated note will be. This capability promises to be a valuable tool in documenting and sharing wine experiences, especially in a world where digital communication is increasingly prevalent.

How concerned should we be as wine lovers? If the input is truly human, is the tasting note still genuine? If not, is this merely exposing an underlying issue: tasting notes have become pompous and embellished, driving the everyday wine lover into feeling like they need LLM assistance to portray their palate?

Maybe "This wine tastes f---ing good" is all we ever needed.

Note: All text in appearing in red italics was generated using an LLM and was not edited by me.