Everyone is talking about artificial intelligence and its benefits, but not everything labeled as AI is actually AI. Some companies are slapping that AI label on technology that’s nothing more than a self-contained module that gathers data without providing any true analysis.
For retailers, tapping into real AI rather than fake AI can make all the difference in sell-through rates and inventory levels. Here’s what retailers need to know about fake AI and how to spot it.
Fake AI is an island
Fake AI is merely an island consisting of a single data set that’s not connected to other data sets. As a result, it can’t provide the complete analysis necessary to gain a true picture of your retail operations. On the other hand, real AI platforms tap into multiple data sets and are capable of analyzing those data sets cohesively, providing the information retailers need to accurately predict how much product in what sizes and colors they will need in each store.
This is where a discussion of terms like “AI module” and “AI native” comes in. AI module implies that the systems, products or services function with AI, but AI isn’t intrinsically tied to the basic function. AI modules use the technology to add features or capabilities and enhance the user experience, similar to how smartphones incorporate AI cameras to improve photography with their devices.
However, AI-native platforms put the technology at the core of product development, decision making or business processes. Instead of AI being simply an add-on technology, it’s a primary, foundational component of the platform. AI-native platforms have the technology as a central element that drives value from their very foundation.
How to spot fake AI
The easiest way to spot fake AI is to notice nuances in the terms used; unfortunately, it’s not always that easy. Sometimes you may have to dig into the technology and gain a better understanding of how it works in order to spot fake AI.
An important sign of fake AI is the use of individual modules, which create islands in the data — islands that can’t communicate with each other because AI is not the central component tying them together. AI modules require scientists to analyze the results of the data gathered. Thus, they are cumbersome and make it impossible for retailers to successfully leverage the results of any information gathered from the AI module.
Another red flag can be the use of the term “forecasting module.” This type of module provides information side by side, similar to a copy/paste of information. Again, use of a module results in an AI island where the information isn’t seamlessly incorporated throughout the system.
It takes very little time to build an AI module, but to embed AI into an entire platform is like inserting it into the platform’s DNA. Like the cells in our body, each AI cell has a different purpose and is integrated into every component of the platform, built into every single feature.
Real AI at work
Believe it or not, up to 60% of retail merchandise ends up sitting on the shelf or gets stocked out due to poor assortment accuracy. Real AI provides much greater accuracy and becomes better over time with more data and adjustments made by people who understand the nuances of their industry.
For example, size profile is critical in assortment planning, allocation and replenishment. When building an assortment plan, the last step involves determining the size quantities. You need to have the right size distribution to keep products from sitting on the shelves for too long or selling out too quickly.
True AI will continuously fine-tune every store location for every product category, determining the ideal size curve each store should carry, and then applies that to the retailer’s style-color-base assortment plan to get down to the style, color, size, and purchase order. True AI based size profiling is embedded in pre-season assortment creation and store allocation recommendations.
Another example of true AI at work is in the open-to-buy analysis, which predicts when retailers will run out of stock. This is a form of forecasting, but it isn’t just a module providing side-by-side information. It’s real analysis from an AI that has access to the entire library of information.
When AI works
With all the marketing around AI right now, it’s important to mention that when AI works, users don’t really need to know. They’ll just use the platform and over time, gradually realize that the data it provides is highly accurate.
On the other hand, AI modules will generate loads of information that requires you to analyze it yourself, potentially leading to disaster if you misinterpret that data.
Real AI is like the brain running in the background of the platform. It just works without any fuss or need to emphasize that AI is involved.
