Get ready for a future in which machines program themselves, the cloud is atomized into a network of intelligent devices and data ownership is a source of competitive advantage.
That’s the world James Kobielus envisions as he assumes his new role as lead analyst for data science, deep learning and application development at research firm Wikibon, a sister company of SiliconANGLE.
Kobielus’ varied background includes roles as a telecommunications guru, big data analyst at Forrester Research Inc. and, most recently, big data evangelist at IBM. He’s also well known as a prolific blogger, speaker and frequent guest on SiliconANGLE’s video studio theCUBE. In a wide-ranging discussion at Wikibon’s Marlborough, Massachusetts, headquarters, Kobielus asserted that the third age of computing is upon us.
The first was defined by logic hard-coded into machines. The second, which is winding down, has humans defining logic and program code. The third age will see machines inferring program logic from the data itself, with the role of humans shifting to user experience and infrastructure orchestration.
Data will assume a more central role in nearly everything organizations do. Data ownership is already becoming a dominant barrier to entry in many markets, Kobielus said, citing Uber Technologies Inc. and Google as two examples.
The tools and skills to work with data will become part of the fabric of any successful organization, although the need for skilled data scientists won’t abate, he predicted. A new generation of professionals is already adjusting to this reality. “More and more of the capabilities to do what we now call data science are available to everybody,” Kobielus said, “and lots of young people now know backwards and forwards how to work with data.”
The impact of machine learning will be enormous over the next few years as the first elements of machine-generated programming are put in place. Kobielus cited the voice recognition capabilities in Apple Inc.’s Siri, Google Inc.’s Voice and the Amazon.com Inc.’s Echo as early examples. “Voice recognition has gotten scary good across the board,” he said. “It’s becoming the coin of the realm around for interfaces, and it demands machine learning.”
Autonomous vehicle networks will be another prime driver of innovation as self-learning networks coordinate the movement of vehicles without human intervention. “The engineering culture [in Detroit] has shifted toward a completely digital focus,” said Kobielus, who grew up in the Detroit suburbs. “The old-line industries are becoming driven by process automation, and much of that is driven by deep learning and the industrial Internet.”
Into the fog
As machines become more intelligent, they’ll also become more distributed. Cloud computing as we now know it will give way to loosely federated networks of connected intelligent devices that Cisco Systems Inc. has dubbed “fog computing.”
Kobielus calls it a wholesale shift from the past. “The whole notion of the data center is going away,” he said. What will replace it is “a radically decentralized cloud fabric where much of the application logic is not hard-coded but inferred from the data in real time.”
Applications will be embedded in application containers so they can move smoothly across a fabric of devices ranging from servers to microsensors embedded in the human body. “Server farms will be atomized to edge devices and containers that are orchestrated over complex fabrics like Kubernetes,” he said. “The logic of the orchestration will be laid down by humans.”
So what’s the role of humans in all this? Kobielus doesn’t downplay the risk of structural disruption as jobs are automated, but he believes humans will be needed at a higher level to figure out applications of data. Hiring managers should focus less on specific skills than on adaptability and willingness to learn.
“Hire smart, creative people who understand that inferring the application of logic directly from data is the way of the future,” he suggested. “Less and less coding will need to be done.”
Kobielus also joined Wikibon Research Analyst Stu Miniman and Chief Analyst David Vellante for a video discussion of some of these issues: