So much can happen in a year, and it seems clear that, if anything, the pace of AI adoption has accelerated even more.
A year ago, we released the results of a survey conducted to determine the state of AI. What we found at the time was a surprising level of momentum behind AI adoption: Not only was AI being adopted at a high rate, but respondents were supporting their AI efforts with increasingly mature processes and infrastructure.
Things are unfolding more quickly than many may have predicted – if business and technology leaders were previously taking a cautious, wait-and-see approach, today they’re all in. Just ask Lisa Spelman, VP of Intel’s Data Center Group. From Intel’s vantage point, “AI is the fastest-growing workload within the data center,” Lisa told me. “If you add in the edge, it only gets bigger. This is only going to continue because AI is basically being built into all applications these days.”
This is great news. AI technology has made huge strides in a short amount of time and is ready for broader adoption. But as companies accelerate their AI efforts, they need to take extra care, because as any police officer will tell you, even small potholes can cause problems for vehicles traveling at high speeds. For example, in AI, what appears to be a minor process or data issue can quickly escalate into an issue with wider repercussions. This isn’t a reason to slow down with AI. Just think of it as being more “road-aware.”
Fortunately, the experiences of others over the past year and before can serve as a beacon for your own AI efforts. Among the customers we’ve worked most closely with, the following issues are some of those they have encountered frequently.
SAS, Intel and Accenture, working with Forbes Insights, surveyed business leaders and interviewed thought leaders around the world to identify early adopters and uncover their emerging best practices for AI. Valuable takeaways from the study include insights on ethics, oversight, process implications and the likely need to shift education and training to prepare for realizing the full potential of AI.
Governance to support growth
When I asked Peter Guerra, Accenture’s chief data scientist for North America, whether he’s seen the same level of heightened activity on AI among Accenture’s clients, he agreed. He went on to say that governance, while not a new issue, has emerged as a primary focal point for the companies they’re working with as they take their AI initiatives to the next level. “We’ve seen clients adopting more AI models across their organizations,” Peter says. “So, governance, model operations, and the whole issue of scale all loom large over that expansion. As a result, we’re seeing technology adoption in closely related areas such as cloud data and model management tools.”
Legacy analytics capabilities figure heavily in the AI governance conversation, given the numerous areas of overlap between AI and analytics. The companies that are aggressively expanding their AI capabilities are generally the types of organizations that have lots of legacy analytics capabilities in place. And those capabilities must be modernized and brought into the AI era. That means reviewing every analytics model and determining whether its underlying assumptions are still true. For companies that have been successfully using the same models for years, this may seem like overkill. But AI introduces new data infrastructures, techniques, and data types – not to mention new desired outcomes. Your analytics capabilities need to operate within the same governance structure in order to ensure that the future growth of these systems unfolds smoothly.
AI technology has made huge strides in a short amount of time and is ready for broader adoption. But as companies accelerate their AI efforts, they need to take extra care, because as any police officer will tell you, even small potholes can cause problems for vehicles traveling at high speeds.
AI as transformation accelerator
Technology leaders often report that due to the enormous amount of hype surrounding AI, their business counterparts frequently seem to want to adopt AI for AI’s sake. These conversations often start with “how can we be using AI?” rather than “I have an untapped business opportunity or stubborn problem.” They’re seeing AI work elsewhere, and they want that, too. Reframing these conversations in the context of broader business or technology transformation goals can help. Because today it’s clear that for many of our customers, AI’s highest and best use is as a transformation accelerator. Planning to completely overhaul your drive-through? Rewiring your customer service experience? Creating a more responsive supply chain? These are the types of transformation challenges where we have witnessed a growing, heavy reliance on AI to speed up and improve the process. When it’s time to have the AI conversation, transformation may be the best starting point.
Surprise! Cultural challenges to AI adoption still dominate
For many business leaders, the emergence of AI is like an unwanted, surprise visit from the doctor. “Can we have a look at that data?” “Why does this process work in this way?” “Does this solution work with that solution?” “Have you seen how they’re doing it in this other department?” Imagine being on the receiving end of these questions, and it’s easy to see the role that culture plays in AI adoption.
AI is powerful. It can also prove to be a real challenge to organizations accustomed to doing things a certain way. Perhaps even more important, it tends to shine a bright spotlight on parts of the organization that many have been content to simply leave in the dark. By now it’s standard practice to identify cultural issues as a potential obstacle to technology adoption – or, really, any scenario in which people are required to do things differently. But the past year has shown that if anything, culture plays an even larger role in successful AI adoption. As your organization expands its AI strategies, plan on taking extra care to manage the cultural side of the equation.
- A guide to machine learning algorithms and their applicationsDo you know the difference between supervised and unsupervised learning? How about the difference between decision trees and forests? Or when to use a support vector algorithm? Get all the answers here.
- Nerd in the herd: protecting elephants with data scienceA passionate SAS data scientist uses machine learning to detect tuberculosis in elephants. Find out how her research can help prevent the spread of the disease.
- What do drones, AI and proactive policing have in common?Law enforcement and public safety agencies must wrangle diverse data sets – such as data from drones – in their proactive policing operations. To be most effective, they need modern tools that support AI techniques like machine learning, computer vision and natural language processing.