If you feel your organization is a laggard with artificial intelligence, don’t feel bad — it turns out everyone else is struggling with it too. AI may be the talk of the town these days, but it’s only just getting out of the starting gate at most companies.
The question is, will AI remain in the labs while the industry moves on to the next technology trend, or will it become the revolutionary force some are predicting?
IBM, which released a study of more than 550 executives, which finds plenty of interest in AI, a full commitment to moving it forward, but very cautious progress. Many organizations are reporting they are still in testing mode, and more than half of the executives are still either experimenting or testing on a limited basis around their organizations. One in seven are only at the planning stage.
That’s the challenge with AI and Machine Learning — bringing it out from proofs of concept or labs and scaling it to meet large-scale production requirements.
Where to start? There are several flavors of AI, and organizations are tooling around with as many of these as possible, whether it’s machine learning-based, deep learning-based, natural language-based, or visual learning-based applications. At least 15 percent are using all four forms of AI within their walls, but the majority, 69 percent, are using some combination of pairings of machine learning, deep learning, and visual AI.
The projects need to demonstrate they will provide returns to the business — or else they will forever remain science projects. “Big business issues trump technology issues by 71 percent and collaboration is a challenge everywhere,” according to IBM’s Kim Storin. “This is not surprising when you think about how hard it is to design and scale AI — AI is inherently difficult because it is an iterative process, not a step-by-step process. Building an AI center of gravity requires new rules for technology, data and infrastructure, but it also demands new organizational approaches, new collaboration models and new processes.”
There are ways to guide AI efforts out of the labs and demonstration projects into actual business production environments. While the survey’s authors make the case for on-premises approaches to AI (this is IBM’s study, after all), there are common best practices shared by AI leaders from which everyone can learn:
Collaborate, collaborate, collaborate. Build a cross-functional team from across the enterprise to concentrate on AI opportunities. A center of excellence may be the way to support such an initiative, clear of organizational politics. “Our goal is to leverage machine learning across our entire organization through a center of excellence model,” according to Lynn Calvo VP of emerging data technology for GM Financial, quoted in the report. “One of the biggest things that keeps me up at night is moving from experimentation to production.”
Ensure that AI is a business strategy, not just a technical initiative. A majority of surveyed executives, 85 percent, say their AI efforts are business imperatives. “A focus on AI as a business strategy leads to serious and scalable AI initiatives, and increases the value of an AI investment.”
Design for enterprise scale right from the beginning. This requires an expanded infrastructure, the survey’s authors urge. Thirty-nine percent of executives say they need improved processing power above that of a CPU, and 37 percent cite a need for more compute power. “Accelerated hardware that uses both CPUs and GPUs has become the preferred engine for AI. And productivity improves when employees are empowered to access the data they need and deploy their models that help them achieve insights that lead to better decision making.
