3 traps companies should avoid in their AI journeys

This article was written by Bob Friday, Vice President and Chief Technology Officer of Juniper’s AI-Driven Enterprise Business.

In a recent survey of 700 IT pros around the world, 95% said they believe their companies would benefit from embedding artificial intelligence (AI) into daily operations, products, and services, and 88% want to use AI as much as possible. When was the last time you heard that many people agree on anything?

Yes, AI is all the rage because it is the next step in the evolution of automation in doing tasks on par with human domain experts whether it is driving a car or helping doctors diagnose disease. But make no mistake while we are starting to see the fruits of AI here and there: By and large, the industry and most organizations are still in the early days of AI adoption. And as with any new momentous technology, organizations need to develop an adoption strategy specific to their organization to get the full benefits of AI automation and deep learning technology.

The complication as Gartner put it: “How to make AI a core IT competency still eludes most organizations.”

But failing to learn how to leverage the benefits AI/ML will leave an organization at a competitive disadvantage in terms of customer experience and operational efficiency. So, what’s the way to get there? Here are three common traps that companies should steer clear of as they get themselves AI-ready.

1. Data and Mission vagueness

Great wine requires good grapes and great AI starts with good data, but great AI also needs a clear business ROI. The business benefit ROI and the data needed to automate the domain expert task must be clearly defined at the outset of the project if the AI solution is to deliver real value and continue receiving the resources to grow from pilot to production.

AI ingredients, like algorithms and machine learning, sound very science-y, but business AI projects should never resemble science experiments. The “Shiny New Toy Syndrome” is a real pitfall for AI. To avoid succumbing to it, organizations should tie every AI project to specific business outcomes and know the business outcome question and what task you are trying to do on par with a domain expert.

For example, is the objective of using intelligent automation to relieve IT team members of mundane, routine tasks so they can focus on higher-value activities? Beyond the IT department, is it to help the marketing department gain competitive advantage by delivering more personalized experiences to customers? Is it automating more of the sales process to boost lead volume and close rate?

C-suite leaders would have to be living under a rock at this point not to recognize AI’s potential and the fact that investment is required for AI-ready technology stacks, but they’re going to want to understand how it’s good for the business. Everyone in a company needs to recognize this reality, and ward off any squishiness in an AI project’s reason for being.

2. Lack of AI/ML skills in the company

The AI talent shortage is often cited as one of the tech industry’s toughest challenges. It has even been called a national security threat amid China’s ambitions to become the world leader in AI.

According to O’Reilly’s 2021 AI Adoption in the Enterprise report, which surveyed more than 3,500 business leaders, a lack of skilled people and difficulty hiring tops the list of AI challenges.

To make sure their companies have the talent to fully leverage the benefits of AI/ML they should start both a hiring and training program.

On the hiring side, companies should look for talent beyond the typical data science degree and look at adjacent degrees such as physics, math and self-taught computer science. But hiring talent is not enough for a companies’ strategy to build their AI workforces, especially when they’re competing with behemoths like Amazon and Facebook. Another good solution to consider: If you can’t hire them, train them.

While it’s unreasonable to expect someone to become a data scientist after taking a couple of online Coursera classes. Engineers with Physics, Math and Computer Science backgrounds have the foundation to master data science and deep learning.

Sources of talent may exist inside the organization in unexpected places. Take, for example, the large business intelligence (BI) ecosystems that many companies have. These have talent that is familiar with using Bayesian statistical analysis that is common to most machine learning algorithms.

In making sure they have the right skills to support AI initiatives, it makes sense for companies to re-train existing employees as much as possible in addition to having an AI/ML hiring strategy. Companies need to get creative in pinpointing those employees and AI/ML talent.

3. Building rather than buying

I’ve seen companies get bogged down by trying to build their own AI tools and solutions from scratch rather than buying them or leveraging open source. The algorithms being used to develop AI solutions are fast evolving and companies should look to partner with vendors in their industry who are leading the AI wave. Unless it happens to be one of the company’s core competencies, building AI solutions is usually an overreach. Why reinvent the wheel when you can buy one of the many commercial AI tools on the market?

Deloitte’s most recent State of the AI in the Enterprise report, which surveyed 2,737 IT and line-of-business executives worldwide, found that “seasoned” and “skilled” AI adopters are more likely than “starters” to buy the AI systems they need.

“This suggests that many organizations may go through a period of internal learning and experimentation before they know what’s necessary and then seek it from the market,” the report said.

Companies that avoid these three traps will have a much easier time accelerating their AI adoption and enjoying the benefits of revenue growth, lower operating costs, and improved customer experience.

Bob Friday is Vice President and Chief Technology Officer of Juniper’s AI-Driven Enterprise Business.