Data moves the world
Artificial intelligence has a trust problem. At the root of this trust problem is the data that powers AI. According to a Salesforce survey of 6,000 knowledge workers worldwide, nearly six in 10 AI users say they have a hard time getting what they want from AI, and more than half, 54%, claim they don't trust the data used to train current AI systems. Three in four of those who don't trust the data that trains AI also believe that AI lacks the information it needs to be useful.
AI innovation doesn't happen in a vacuum, or by business or technology teams working on models and algorithms: it requires a robust and well-validated data foundation.
As the complexity of today's AI systems and applications grows, industry leaders are raising concerns about the feasibility and reliability of the AI being put into them. “AI is only as good as the data that supports it,” says Sean Knapp, founder and CEO of Ascend.io. “Business leaders and experts need to understand that just because AI gives us an answer doesn't mean it's accurate.”
In this regard, Knapp adds that “AI's enormous innovative potential cannot be realized by simply pushing it hard,” adding that “data development is often siloed and slow, fraught with delays, disconnection and disillusionment.”
After all, a data-driven business is an AI-driven business, and there is no longer any distinction between the two definitions. “Without a data-driven focus, companies cannot compete,” says Sharad Varshney, CEO of OvalEdge.
The problem, Knapp says, is that “many organizations are still not familiar with data for basic business intelligence tasks, let alone AI.” What's needed is “clean data from advanced data pipelines. Achieving operational efficiencies, improving customer experiences, and creating innovative products depends on building systems that quickly identify and reliably generate the datasets you need.”
“If you adopt AI-enabled data management, analytics and governance technologies from day one, you'll be in an enviable position,” Varshney points out.
First and foremost, organizations need to understand that without the right data, AI initiatives won't get past the starting point. “Many business professionals are eager to jump right into analyzing and leveraging AI models without thinking about building a solid foundation of data,” says Jonathan Bruce, vice president at Alation.
“You need to slow down to speed up,” Bruce continues. “While there are benefits to rapidly adopting AI, the organizations that will emerge strongest from the AI revolution will be those that have invested in a solid foundation of trusted, governed datasets on which to base their AI initiatives. Packaging in trust allows users to understand the provenance and lineage of supporting data, and empowers them to apply those models at the speed of business.”
To keep up with the speed of business and drive innovation forward, “data and AI are inextricably linked, and companies need data to train their AI solutions,” says Ram Chakravarti, CTO at BMC. “AI not only makes new data available, but also helps with better analysis and identifying patterns and anomalies. Moreover, AI can automate routine tasks, freeing up employees' time to focus on new business ideas and structures.”
High-quality data is essential, says Chakravarti. “For AI to be valuable, it needs to be trained on high-quality datasets, and data quality is just as important as data quantity,” he says. At the same time, he adds, “without AI, organizations will find it hard to derive meaning from large amounts of data.”
