This pic and the reason for this piece came from a recent article posted at a wine media outlet called the Wine Industry Network here is the link: Predicting Palates: Can Artificial Intelligence Improve Wine Buying? – Wine Industry Advisor. I typically enjoy this website. The author piqued my interest and I decided to dive deeper. I felt the article needed more perspective. This topic really requires an effort to validate the solution, to have relevance.
Past and Future Attempts at Wine AI
This particular software branded as “Tasty” by name is similar in concept to others like “Quini” (if not in process), all have one very major flaw… Very few consumers have the trained palate and sensory awareness required to describe what they are tasting and how they perceive the wine components. I have been approached by software developers before and none had an appreciation for a trained palate and what it brings. These techno driven business models providing wine related services are all smoke and mirrors. Sounds great, until you dive a little deeper and find the missing piece. The idea that a straight chemical analysis, or even an analyzed database of wine tasting notes could provide any real insight into how wine is perceived by the human palate is misguided. Software can certainly accumulate, organize and label wine data, but how does that data have any relevance – unless it is filtered through human perception?
What Is Missing?
I have put some thought into this before and shared those ideas with other software developers in the past. The baseline in the software database must be established with trained professionals first – and not with WSET 4, MW, or CSW certified pros. The calibration has to be with Master Somms (MS) who evaluate consumer palates in restaurants on a daily basis. Once the underlying premise is established and the work is done to connect the human palate (sensory experience) to the chemical evaluation, the concept still requires a short consumer educational program to provide a shared vocabulary. That vocabulary is the vehicle for the shared human wine experience. Which words are chosen to describe each individual sensory experience? Is there a common understanding of what flavors/components they represent? I have yet to be introduced to software that attempts to address these concerns. Without that effort, the whole endeavor is a major waste of time and money.
Am I Biased?
I don’t think so. Regardless of my Somm training, I make my primary living in the technology and automation field. Perhaps that background gives me a little authority here. I most definitely believe the goal of these “AI” software products is achievable, it just requires more trade awareness to succeed.