AI is All Natural: What to Do About Ambiguous Labels

As the all-natural eating movement went mainstream in the 90’s, food and beverage companies rushed to slap the “All Natural” label on their products to meet market demand.

For many of those companies, the change involved nothing more than updating the packaging. It worked. Neither the public nor the USDA—the food regulation agency—had a concrete idea of what “All Natural” meant.

For a while it wasn’t clear who’s fake and who’s genuine. To rise above the noise, manufacturers making a real effort to produce healthy food needed to say more than just “All Natural.” More specific abels such as “Organic” and “Free-range” entered the food marketing lexicon, and companies began positioning themselves as “transparent” and “honest,” filling significant space on their labels with origin stories and mission statements.

Today, the distance between fake and genuine has spread, and it’s easier to discern between the two. Consumers don’t need to become food scientists to tell them apart, they need only to walk into the right store. (Nevertheless, putting an “All Natural” label on food is still a profitable gimmick, since the USDA never defined “Natural” beyond broad terms, and several court cases allowed the practice.)

Now let’s talk about software…

There is enormous demand for software with “AI.”” Multiple founders and executives of software startups tell me that their prospective buyers, current customers, lead investors, and market analysts are all pushing for some AI functionality.

In the rush to meet market demand, some software companies are finding ways to improve or supplement their core products with computer vision, language processing, anomaly detection, recommendation engines, automated decision-making, deterministic predictions, and other genuine implementations of AI.

Other companies just slap the “AI” label on their sales and marketing materials and reap the benefits of an ignorant market and an ambiguous term. (I will refrain from pointing to examples, as easy as it is to find them.)

Software companies that are making a real effort to solve customers’ challenges with AI must find a way to rise above the noise. Here are two ways:

• Determine and share how the new, AI-powered functionality will help software buyers achieve their desired outcomes. The “AI” label alone might convert some early adopters for a while longer, but scrupulous stakeholders will always want to know how they stand to benefit.
• Provide transparency in how the AI functionality works and how it was made. Use more specific language, show system diagrams, spotlight data scientists and machine learning engineers, offer to be a case study for the AI/ML tools they use, and talk about the process.

The “AI” label is not going away. On the contrary, it might replace the word “software” as AI/ML melds into the core of most software products. (As in: “We need to buy a marketing AI.” “We sell AI for IT teams.” “Should we build our own sales AI?”) Nor is there a government agency to save us from dishonest use of the term.

Therefore, the companies making a real effort of improving the lives of their customers with AI have two options: They can waste energy decrying the buzzword and trying to convince buyers, investors, and analysts why they’re wrong and naive, while the fakers keep taking market share… Or they can get more customers by embracing the language the market has chosen to use, and by helping buyers expand their vocabulary to better discern between real and fake.

Note: During editing of this post, I found another article with the same “AI” = “All Natural” analogy. I decided to publish my perspective anyway, despite not being first-to-market.