Analysts predict that by 2020, artificial intelligence technologies will be in almost every new software and service...
release. And if they're not actually in them, technology vendors will probably use smoke and mirrors marketing tactics to make users believe they are.
Many tech vendors already shoehorn the AI label into the marketing of every new piece of software they develop, and it's causing confusion in the market. To muddle things further, major software vendors accuse their competitors of egregious mislabeling, even when the products in question truly do include artificial intelligence technologies.
AI mischaracterization is one of the three major problems in the AI market, as highlighted by Gartner recently. More than 1,000 vendors with applications and platforms describe themselves as artificial intelligence products vendors, or say they employ AI in their products, according to the research firm. It's a practice Gartner calls "AI washing" -- similar to the cloudwashing and greenwashing, which have become prevalent over the years as businesses overexaggerate their association to cloud computing and environmentalism.
AI goes beyond machine learning
When a technology is labelled AI, the vendor must provide information that makes it clear how AI is used as a differentiator and what problems it solves that can't be solved by other technologies, explained Jim Hare, a research VP at Gartner, who focuses on analytics and data science.
Jim Hareresearch VP, Gartner
"You have to go in with the assumption that it isn't AI, and the vendor has to prove otherwise," Hare said. "It's like the big data era -- where all the vendors say they have big data -- but on steroids."
"What I'm seeing is that anything typically called machine learning is now being labelled AI, when in reality it is weak or narrow AI, and it solves a specific problem," he said.
IT buyers must hold the vendor accountable for its claims by asking how it defines AI and requesting information about what's under the hood, Hare said. Customers need to know what makes the product superior to what is already available, with support from customer case studies. Also, Hare urges IT buyers to demand a demonstration of artificial intelligence products using their own data to see them in action solving a business problem they have.
Beyond that, a vendor must share with customers the AI techniques it uses or plans to use in the product and their strategy for keeping up with the quickly changing AI market, Hare said.
The second problem Gartner highlights is that machine learning can address many of the problems businesses need to solve. The buzz around more complicated types of AI, such as deep learning, gets so much hype that businesses overlook simpler approaches.
"Many companies say to me, 'I need an AI strategy' and [after hearing their business problem] I say, 'No you don't,'" Hare said.
Really, what you need to look for is a solution to a problem you have, and if machine learning does it, great," Hare said. "If you need deep learning because the problem is too gnarly for classic ML, and you need neural networks -- that's what you look for."
Weak AI vs. strong AI
Broadly, there are two types of AI:
Weak AI is pervasive today in the form of chatbots, which serve a specified purpose.
Strong AI tools go much further; these tools come up with solutions to problems on their own, through massive amounts of data and cognitive computing capabilities.
Don't use AI when BI works fine
When to use AI versus BI tools was the focus of a spring TDWI Accelerate presentation led by Jana Eggers, CEO of Nara Logics, a Cambridge, Mass., company, that describes its "synaptic intelligence" approach to AI as the combination of neuroscience and computer science.
BI tools use data to provide insights through reporting, visualization and data analysis, and people use that information to answer their questions. Artificial intelligence differs in that it's capable of essentially coming up with solutions to problems on its own, using data and calculations.
Companies that want to answer a specific question or problem should use business analytics tools. If you don't know the question to ask, use AI to explore data openly, and be willing to consider the answers from many different directions, she said. This may involve having outside and inside experts comb through the results, perform A/B testing, or even outsource via platforms such as Amazon's Mechanical Turk.
With an AI project, you know your objectives and what you are trying to do, but you are open to finding new ways to get there, Eggers said.
AI isn't easy
A third issue plaguing AI is that companies don't have the skills on staff to evaluate, build and deploy it, according to Gartner. Over 50% of respondents to Gartner's 2017 AI development strategies survey said the lack of necessary staff skills was the top challenge to AI adoption. That statistic appears to coincide with the data scientist supply and demand problem.
Companies surveyed said they are seeking artificial intelligence products that can improve decision-making and process automation, and most prefer to buy one of the many packaged AI tools rather than build one themselves. Which brings IT buyers back to the first problem of AI washing; it's difficult to know which artificial intelligence products truly deliver AI capabilities, and which ones are mislabeled.
After determining a prepackaged AI tool provides enough differentiation to be worth the investment, IT buyers must be clear on what is required to manage it, Hare said; what human services are needed to change code and maintain models over the long term? Is it hosted in a cloud service and managed by the vendor, or does the company need knowledgeable staff to keep it running?
"It's one thing to get it deployed, but who steps in to tweak and train models over time?" he said. "[IBM] Watson, for example, requires a lot of work to stand up and you need to focus the model to solve a specific problem and feed it a lot of data to solve that problem."
Companies must also understand the data and compute requirements to run the AI tool, he added; GPUs may be required and that could add significant costs to the project. And cutting-edge AI systems require lots and lots of data. Storing that data also adds to the project cost.
A look at deep learning vs. machine learning
The power, promise and controversy of AI
AI apps need a reality check