Recently, I was chatting on LinkedIn with my colleagues on the topic of the future of AI, more specifically, data. Many pundits have commented that data is the fuel that feeds the AI fire. Or it’s the electricity of the 21st century. Or it’s another metaphor that essentially means AI is meaningless without data. Pick your metaphor, pick your poison. 

The problem is most of the hottest AI applications are powered by internet data. As these AI models dominate the internet, they will be fueled by the outputs they previously generated. This will lead to “model collapse,” the gradual degradation in the output of a generative AI model trained on synthetic data. Talk about a self-fulfilling prophecy! 

This data source will eventually deplete its usefulness. Instead of mining content for AI applications, enterprises should seek to leverage the rich data sets locked away in their things. According to Exploding Topics, “there are well over 14 billion connected IoT devices around the globe. By 2030, that number is expected to reach more than 25 billion.” 

Consider this insightful exchange from Ernst & Young: 

The rise of AI will tend to favor large, data-rich companies over smaller upstarts, Cisco CEO Chuck Robbins told me Monday, just after the close of the company’s $28 billion acquisition of cybersecurity and analytics company Splunk.

“This is a revolution that actually favors incumbents with large datasets, and that’s unique. A lot of times in the past, the big technology waves have allowed smaller companies to compete and disrupt markets. That’s still a risk, but large companies with large datasets that use AI effectively are going to increase their competitive differentiation,” Robbins said.

I couldn’t agree more. Buildings, machinery, valves, pumps, vehicles, video cameras, and so much more are teeming with useful information. They will reduce costs, improve efficiencies, and provide better service when leveraged correctly.

Historically, that data has been nearly impossible to access. Most manufacturing data, by some estimates, the second-largest data set in the world, is locked on-premise, behind firewalls, or completely disconnected from networks, making it nearly impossible to stream.

That is no longer the case. IoT platforms paired with Edge & Edge AI have been unlocking that data, giving companies unprecedented insights into this rich data set. It works by retrieving and sending data from any device via any protocol, formatting and normalizing data into a common structure, streaming it back to data lakes, and processing and acting upon it in real time.

Action is the key verb that makes Edge AI so powerful. Instead of just contributing more data to a data lake without generating insight to improve the business, the process looks like this: 

As data scientists understand the data lake’s information, those correlations are put to work back in the field where an operator immediately knows to reduce the pressure on a drill bit at a fracking site or where a maintenance worker immediately knows that the optimal way to complete their next task is with a specific tool configured with certain settings. 

The future of AI will be powered by machine data. The companies that can harness the massive quantities of machine data will be the real winners in this AI race.