While artificial intelligence is still a work in progress, it presents a massive opportunity for CIOs to reinvent how organizations think about the CIO role.

To say a CIO’s job is difficult is a gross understatement. Carrying out routine IT duties and deploying cutting-edge innovations makes for a tough balancing act. Toss in the fact that every CIO in the world is now under pressure to produce a game-changing AI strategy and you’ve got what seems like a recipe for career derailment.

But I believe AI is different — and presents a massive opportunity for CIOs to reinvent how organizations think about their role.

AI—and more specifically, deep learning—has now reached a tipping point thanks to more data, greater computing power, and much-improved algorithms that are swiftly jumping from academia into the enterprise. Investment in AI capabilities is growing as well, according to IDC, with global spending slated to grow from $24 billion in 2018 to more than $77 billion in 2022.

But there’s a learning curve with any wave of technology. So where should the smart CIO start?

Understand how AI is different

The most essential idea for a CIO to adopt is that AI is not just another tactical technology (like, say, file-sharing or survey tools). It’s a strategic technology that is not only capable of improving existing operations, but also opening new lines of business.

Unlike other machine learning algorithms, AI is a hungry beast that demands vast amounts of well-labeled information. In return, the best AI provides better and less expensive predictions that can move your business in surprising directions.

It’s easy at this point to think of AI as the panacea for all that ails your organization. But that would be a mistake, because AI can also expose your hidden weaknesses and create a few new ones.

AI differs from more traditional machine learning algorithms in one important way: It sifts relentlessly through all data and assigns weights to the most relevant aspects. While this is powerful, it can expose unseen biases in your data, which can mislead your teams and unfairly discriminate against your customers.

Start simple, win fast

Given the pressures that CIOs face, it’s tempting to embark on a comprehensive AI project. But for many, that should be avoided.

Since AI itself is in a constant state of evolution, your approach should instead focus on experimentation and learning; this lets you better align your growing understanding of AI’s  capabilities with your business needs. If you follow this path, you’ll have better control over AI investment scope and results.

The best way to get started is by directing efforts at opportunities ripe for optimization. Your first AI projects will have the greatest chance of success if/when you:

• Experiment on small problems with high ROI
• Start with existing software libraries like Tensorflow (Google) and PyTorch (Facebook)—or a robust data management platform like Domo, where AI is baked in—instead of building your own model
• Use your best available, validated, highest-quality data
• Integrate AI into an existing business process where disruption will be minimal
• Focus on productivity by automating well-defined and highly repetitive decisions

You can help others across the organization get more value from their data with quick wins. And building confidence will lead to larger projects and higher visibility.

Think like a CEO

As mentioned above, AI is a strategic tool. As such, it requires strategic thinking. As aspirations begin rising, the path to greater AI adoption will be filled with challenges. A CIO who can think like a CEO will be proactive in understanding and addressing misaligned department strategies, shifting priorities, and the politics that dampen collaboration.

One way to rise above the inevitable contentiousness and doubt is to demonstrate unwavering focus on finding new ways to improve and support the business. This is where a CEO-like understanding of the business is critical. You need to credibly pitch AI as a potential answer to business questions across departments.

It helps to see the potential by thinking in terms of an “If we can A, then we can B” formula. For instance, if you’re in customer service, you might say, “If we can sort customer emails by sentiment, then we can make customers happier by addressing issues more quickly.” If you’re in sales, you might say, “If we can better score leads, then we can make our sales team more effective.”

By focusing on solutions, you’ll be able to reinvent how other leaders in the organization think about you. With AI, you can provide value on top of every data source (marketing, sales, operations, finance, etc.) in the company.

Approach AI, well, intelligently

Great AI thinker J.C.R. Licklider proposed a formula for human and machine cooperation decades ago that has begun playing out in interesting ways across everything from chess competitions to new startups:

Average Human + Machine + New Process > Smart Human + Machine + Old Process

This formula represents a departure from the “start simple, win fast” mentality in one significant way: You must actively develop new business processes—and ones that involve delegating tasks based on strengths. Machines handle the routine tasks, opening the way for humans to discover greater insights and make better decisions.

As a CIO, you might not have total authority over all of this, but you can certainly help support and communicate the big idea.

Embrace the AI wave

The tech industry has a long history of flash-in-the-pan buzzwords that went nowhere. AI isn’t one of them. Don’t be the last one to jump in and embrace what AI can do for your organization. Instead, make sure you’re managing your company’s expectations of the value AI can provide. By taking a strategic approach to this new wave, you can manage your investment while still benefiting from the power of the next big thing.