Only 21 percent of small businesses have implemented artificial intelligence (AI)-based solutions, according to a report from Bluewolf (an IBM company). The AI Investment Gap Survey polled 177 decision makers around the globe to determine whether they had adopted AI and machine learning (ML) yet, and the depth of their understanding of these technologies. Although 33 percent of small businesses planned to invest in AI within the next 12 months (bringing the total number of AI adopters next year to 54 percent), the total is still lower than that of large companies. Plus, 30 percent of large companies have already invested in AI while 44 percent plan to begin investing within the next 12 months. This brings the total to 74 percent, or 20 percent more than the total of small businesses.
Vanessa Thompson, Senior Vice President of Customer Experience Insights at Bluewolf, said a knowledge gap exists between companies that have adopted AI tools and those that aren't planning to adopt such tools. She calls this gulf "the AI Investment Gap" and describes it as a "discrepancy between C-level executives who understand AI and those who have yet to deploy it into their business," according to a written statement.
Because Bluewolf sells AI tools, it would behoove them to suggest that the only reason people don't buy AI tools is because they don't know about them. To check Thompson's claim, I spoke with Brandon Purcell, Senior Analyst of Customer Insights at Forrester Research, about what, if any, other issues might exist to cause the gap between those who have adopted AI and those who haven't. Purcell and Forrester Research have conducted their own similar studies about AI adoption. Although his overall numbers are similar to IBM's—51 percent of companies have adopted or are expanding AI, and 20 percent say they plan to adopt within the next 12 months—Purcell came up with a couple of other compelling reasons why small businesses might be behind the curve of AI adoption.
The Cost of AI
Purcell referenced investment constraints as a major factor, especially "as it relates to skill set. Small businesses don't have the resources to hire data scientists," he said. These are the workers who will extract insights from the data being pushed into and out of enterprise software.
They'll also be the ones who determine whether the AI is accurately reading your data and taking actions based off of its own intelligence. The average salary for a data scientist is $113,436 per year, according to Glassdoor, which is (in the grand scheme of rich) just slightly less than the average salary of an American CEO ($166,000, according to PayScale). So, if you're a small business CEO who is operating on razor-thin margins and you don't want to cut your own salary, then it would be difficult to rationalize spending six figures on a data scientist—and spending money on a software system that can turn data in AI.
But it's not just the money involved that prohibits smaller companies from investing in AI-driven software. "On a related note, there's a data factor," said Purcell. "AI flourishes when you have large amounts of data. Small businesses don't have as much data to do that."
Think of it like this: You know how Facebook knows which friends to tag when you post a photo? That's because Facebook has been gathering information from all of your previously tagged posts. You ever watch a movie that Netflix recommended to you? Netflix knew to recommend that movie based on your previous selections. Facebook and Netflix are able to make these recommendations based on ML, which is the first cousin of AI. Although they're similar, both terms are often used interchangeably (and incorrectly).
Here's the basic difference between the terms: ML systems use intelligence to improve performance by offering you recommendations and ways to streamline processes, whereas systems that utilize AI give autonomy to the software to carry out tasks and make decisions without human oversight. ML is Netflix making movie recommendations while AI is a car driving you to work while you take a nap in the backseat. As a small business that is just starting to generate data, the advantages of AI will be miniscule compared to what a Fortune 500 company might see when they turn on their AI software.
Is Bluewolf Wrong?
So, was Bluewolf fed poor information in their survey? Do small businesses know about AI but they just don't have the money or data to get excited about it? Purcell doesn't think Bluewolf's research is wrong. In fact, he credits IBM Watson as the creator of cognitive computing, the umbrella term that encompasses AI, ML, and other applications that mimic the human brain.
"They spent a lot of money to create that category, but they have big competitors in the space: Google, Amazon, Facebook, Microsoft," Purcell said. "Those companies are also sitting on massive amounts of data used to train AI systems. The Hollywood definition of AI is the sentient robot. We haven't used that yet. But, when it comes to implementing AI at the enterprise level for practical AI, IBM is excelling at creating those tools."
Misconceptions about Hollywood, AI, and robots murdering us in our sleep are a likely reason why small businesses have shied away from learning more about AI tools. If you're a t-shirt vendor in Oklahoma, then what good is an autonomous car or a future-robot armed with a laser gun? However, when taken in its lesser-known context, Purcell and Thompson see practical use cases for small businesses—use cases about which small businesses haven't been educated yet.
With something that Thompson and Bluewolf refer to as "augmented intelligence," small businesses don't necessarily need the data expertise or the trove of information to take advantage of AI. Bluewolf defines augmented intelligence as the ability for apps to reason, infer, and extract ideas, even with unstructured data sets, such as language and imagery. Even at the beginning of a company's data collection, augmented intelligence solutions are able to learn as they go, regardless of how little information is being fed into the system.
"Augmented intelligence helps end users predict what to do next by giving them a profile of what their customers need," said Thompson. "We see augmented as a way to make AI a reality for companies of any size."
This includes things such as combining external and internal data to pad the knowledge that the augmented intelligence technology is using to make business decisions. For example, by combining external local shopping patterns and weather data with proprietary, customer shopping pattern data, e-commerce companies can deliver hyper-personalized campaigns. In this scenario, a data scientist would be helpful but not necessary, and a trove of customer data would make the campaign even more powerful. But it wouldn't stop the campaign from being more powerful than it would have been without the combination of internal and external data sources.