If we’re not careful, artificial intelligence is in danger of becoming the tomato of the technology world.
Walk into almost any grocery store, for example, and you’ll find tomatoes near vegetables like broccoli and cauliflower. Many consumers may or may not be aware that, botanically speaking, tomatoes are fruits. A tax issue in the U.S. dating back to the late 19th century led to the industry classifying them as vegetables.
Maybe that distinction doesn’t matter for tomatoes given it’s easy to add them to a recipe, no matter where you find them in the produce aisle.
But when you’re embracing technology like A.I. based on the belief it will make a material difference to your business’ growth and profitability, the labels need to be right.
According to the results of a study published recently by MMC, 40% of European startups describing themselves as A.I.-based companies aren’t using the technology the way that experts would define it.
“In 40% of cases, we could find no mention or evidence of A.I.,” MMC head of research David Kelnar told Forbes, adding that it means “companies that people assume and think are A.I. companies are probably not.
If these companies aren’t leveraging A.I., why are they claiming to do so?
One of the biggest culprits is the misunderstanding between predictive analytics based on statistical modeling that hasn’t changed much in 20 years and “true” A.I. (Note: Here’s a simple explanation of A.I. versus predictive analytics.)
Many of the most talked-about A.I. use cases, meanwhile, have come out of academia. By its nature, academia tends to focus on simpler areas and concepts than what goes on within a large, complex enterprise like a grocery chain or insurance firm.
As well, the technology sector is already complex enough, and everyone from analysts, consultants, and the media often look for easy ways to cluster technologies that aren’t that similar.
As an investor, reporter, analyst or company, it is important to understand predictive analytics and A.I. are different. If a company is mostly using predictive analytics, it is not leveraging A.I.
To separate the A.I. wheat from the chaff, so to speak, here are three things to keep in mind about what makes reinforcement learning a true form of A.I. that you can trust:
Stability and control
The idea of autonomous vehicles may still provoke controversy and debate, but one thing is clear: no one is ever going to step into one if it zig-zags wildly between lanes of traffic.
Instead, the reinforcement learning that serves as the foundation for developing self-driving cars is designed to sort through massive sets of data about movements, routes, and possible traffic anomalies. This ensures autonomous vehicles behave in a smooth, responsible manner.
The same thing is true of A.I. based on reinforcement learning in areas like retail where billions of possible scenarios are modeled to guide decision-making among merchandisers and other members of the team.
Predictive analytics is a guess based on the past — nothing less, but nothing more. That means the technology can be wrong with dire consequences.
Think of a so-called A.I.-based system that suggests, based on historical data, that a grocery retailer order of large quantities of a product like bananas, but the expected demand doesn’t materialize. No one can afford that kind of mistake.
Contrast that with the myriad points of data that feed into more “what-if” scenarios and models than a human being could possibly imagine. You’ll begin to see why reinforcement learning is the only A.I. worth talking about.
In the old world of traditional IT, experts would say that technology could allow you to do things cheaper, faster or better — as long as you picked two out of those three options. True AI means getting the speed, the cost-effectiveness, and the quality of information, along with a truly game-changing outcome.
We’re not talking about growing the bottom line by 2%. Think about 100% of more profit growth for a grocery retailer.
Don’t be fooled by something described as “A.I” that’s really window dressing. A.I. and predictive analytics is not a case of “tomato/tomatho.” It’s the difference between experimenting with approaches that drive incremental improvements and the kind of positive disruption that will eventually make true A.I. easy to identify.
For insight into how A.I. is changing how retailers approaching price, check out our eBook: Autonomous decisions in retail: how artificial intelligence is reshaping pricing.
To learn more about, how Daisy’s A.I.-powered technology helps retailers and insurance companies drive higher profits and revenue, drop us a note.