In recent years, the vast majority of the enterprises that invested in Artificial Intelligence (AI) capabilities fell into one of two categories: those who used AI applications successfully to improve operations or cut costs and those who were participating in what Goutham Belliappa, vice president of AI engineering at Capgemini North America calls “AI theater:” They implemented AI models “to create some buzz in the marketplace, but they didn’t go through the hard work of tying their AI capabilities to business value,” Belliappa says.

Today, companies stand on the precipice of a new era. “AI is on the cusp of a tremendous economic impact that will disrupt every industry in the same way that software was positioned about thirty years ago,” says Brian Jackson, analyst and research director at Info-Tech Research Group. “AI’s rapidly growing capabilities are being applied to solve problems in far more efficient ways than we were able to do previously.”

[ Read also: AI vs. natural language processing (NLP): What are the differences? ]

How to develop an Artificial Intelligence (AI) strategy 

As a result, forward-looking IT leaders are revisiting and rethinking their AI strategies for the future. Belliappa has been working on revising client AI roadmaps to increase revenues through personalization, dynamic pricing, and the creation of new data-enabled revenue streams.

Meanwhile, data and AI products have matured and become mainstream. “The challenge is integrating these AI and data products into a company’s operations, commerce, or other products,” says Belliappa, who notes that data and AI strategies from just a year or two ago are now outdated.

“If a company is not considering how its strategy should incorporate AI and how AI might disrupt their industry, then it will only be a matter of time until they find themselves playing catchup with another competitor that has done that work,” agrees Jackson of Info-Tech Research Group.

Consider these ten questions to ask now about your AI strategy:

1. How much revenue do we drive from AI or AI-embedded products?

“Many firms are adopting machine-learning capabilities to assist with some business processes; for example, using chatbots to triage incoming customer support cases,” says Jackson. “That’s great and useful but unlocking the real potential of machine learning and the value it can create won’t be harnessed until it’s adapted into the core value proposition of a business.”

[ Check out our primer on 10 key artificial intelligence terms for IT and business leaders: Cheat sheet: AI glossary. ]

2. What role do we see ourselves playing in the AI marketplace?

Enterprises should decide where they fit in terms of the risks and rewards of AI, says Jackson. AI leaders will hire data scientists to create their own AI IP to drive business growth, or even sell AI services to others. Early adopters may not develop their own AI algorithms but will be quick to integrate AI solutions by partnering with AI leaders to drive efficiency and revenue growth. Those with the least risk tolerance will simply want to adopt AI features built into the software and cloud products they already use, Jackson says.

3. What outcomes do we seek?

This may sound obvious. However, many organizations are still pursuing AI for AI’s sake. “Artificial intelligence has gathered such momentum as a concept that many business leaders end up understanding that they need it even before they understand what they need it for,” says Vara Kumar, CTO and cofounder of Whatfix. Kumar advocates conducting a thorough audit of an organization’s technical processes.

“Organizations get tunnel vision in attempting to understand what types of opportunities AI can unlock and then map them into organizational goals, versus starting with the organizational goal first and mapping to how AI can help,” says Sam Babic, senior vice president and CTO at enterprise content management and process management software maker Hyland. “This seems like a nuance, but the latter enables the organization to more quickly focus on the requirements necessary to accomplish the goal versus getting lost in a sea of possibilities.”

[ Want best practices for AI workloads? Get the eBook: Top considerations for building a production-ready AI/ML environment. ]

4. What ethical risks should we be monitoring and mitigating?

“With the automated decision-making that comes with adopting AI comes the risk of ingraining a systemic bias into your operations,” says Jackson. “Consider if the decisions you want AI to make will have an impact on people’s lives and where human judgment must be included in the process. There are many efforts underway around the world to issue guidance on AI ethics.”

As part of this, today’s organizations need to ensure they have diverse teams working on AI initiatives to enable their ongoing improvement and efforts toward zero bias, says Michael Ringman, CIO of Telus International.

5. Do we have the capabilities and infrastructure to deliver on our AI plans?

“Organizations must be realistic about what their AI approach is going to be,” Jackson says. “If they are lacking in IT capabilities, such as cloud infrastructure and data warehousing, then there is no skipping straight to the path of AI leader.”

Shane Nolan, head of technology for foreign direct investment agency IDA Ireland, recommends conducting a capability gap analysis, data preparation, and “building AI solutions around readily available data sources – not aspirational ones.”

Let’s examine five more important questions to ask:

6. Is our AI maturation keeping up with data growth?

To stay ahead of the curve, businesses should set a target to match their decision-making speed to that of anticipated growth in data volumes over the next year, advises Euan Davis of the Cognizant Center for the Future of Work. For instance, IT leaders who expect 30 percent growth in data should set a goal of increasing their organization’s speed of insight by 30 percent. “Anything less will impact the speed of doing business in this fast-changing world,” Davis says.

[ Public data sets can help with training. Read also: 6 misconceptions about AIOps, explained. ]

7. How can we find next-gen AI talent?

AI is not just about technology; it is also about people. “Critical to leveraging the possibilities of AI is hiring talent that can understand the technology and business needs and create solutions, not just build models,” says Ben Pring of the Cognizant Center for the Future of Work.

“Organizations should deeply focus on HR plans (hiring and retention) that prioritize securing the next generation of talent; without it, it will be virtually impossible to keep pace in markets that are being disrupted at light speed,” Pring says.

Having the right people in place, including data engineers and data scientists, is imperative to success “as they will be able to identify and correct small issues before they potentially become big problems,” says Josh Perkins, field CTO at digital platform company AHEAD.

Of course, external hiring alone won’t get you there; it also pays to train up AI talent. Read also: Artificial Intelligence (AI): 4 novel ways to build talent in-house.

8. How will humans and machines interact in our environment?

This will guide plans for everything from enhancing the user experience of internal bots and platforms to upskilling and retraining employees. You’re contemplating how, where, and when to deploy AI to complement employee capabilities, says Ringman of Telus International.

9. Should we create an AI center of excellence?

While it’s beneficial to start small and build momentum, there is often value in developing an AI center of excellence. “In the formative stages of AI adoption, it is good to set up an AI center of excellence where subject matter experts either report directly or through the dotted line,” says Babic of Hyland. “This center of excellence provides focus and dedication to the topic and allows a centralized approach to patterns and practices derived through learning.”

10. How can we better democratize AI?

“Some leaders are surprised to learn that democratizing AI involves more than the process itself,” says David Tareen, director of AI and analytics at SAS. “Often, culture tweaks or an entire cultural change must accompany the process. Leaders can practice transparency and good communication in their democratization initiatives to address concerns, adjust the pace of change, and result in a successful completion of embedding AI and analytics for everyone.”

Nolan of IDA Ireland recommends educating the broader team on the art of the possible. “AI shouldn’t just be the preserve of the IT or software engineering team,” says Nolan, who urges workforce education and awareness building.

[ Get exercises and approaches that make disparate teams stronger. Read the digital transformation ebook: Transformation Takes Practice. ]