Artificial intelligence (AI) is so last year, according to some experts.
Scientists at MIT this week claimed a breakthrough in how human intuition can be added to algorithms. And in a separate, unrelated report, Deloitte Consulting is chastising the business community for not comprehending fully that new, cognitive computing technology should be exploited.
“Artificial intelligence is only the beginning,” researchers write in a Deloitte University Press article about Deloitte's February study.
+ Also on Network World: Using artificial intelligence to teach computers to see +
“Advanced cognitive analytics” is just one of the “fast-evolving” technologies businesses need to get a handle on, they say. A kind of artificial intuition and cognition through algorithms is one part of that machine intelligence (MI). Notably, it’s not AI. MI is more cognitive and mimics humans, the firm explains, while AI is simply a subset of MI.
“To focus on AI is to miss the forest for the trees,” writes Blaise Zerega in a VentureBeat article about the Deloitte report.
MI includes machine learning, deep learning and cognition, among other tools like Robotics Process Automation (RPA), and bots. Deloitte says the time is ripe to latch on to umbrella-term MI and stop single-mindedly concentrating on apparently one-dimensional AI.
It cites reasons that include data growth for making it all possible finally. The consulting firm says collected data doubles every year now, and it will reach 44 zettabytes by 2020.
Faster distributed systems, introduced by better chips and networks, sensors and Internet of Things (IoT)—coupled with those huge swaths of data and “smarter algorithms” that “simulate human thinking”—are other important elements that are going to unleash MI over AI.
It’s those advanced algorithms that are probably the most exciting—and the most different.
Adding human intuition to a machine algorithm
Massachusetts Institute of Technology (MIT) is one organization racing to design them. The school recently said it now knows how to include human intuition in a machine algorithm. That’s a big deal.
It’s going to do it by copying how clever people solve problems, researchers say in an MIT News article.
In recent testing, it asked a sample of brainy MIT students to solve the kinds of issues that planning algorithms are used for—like airline routing.
Problems in that field include how to optimize a fleet of planes so all passengers flying the airline network get to where they want to go, but no plane flies empty and doesn’t visit a city more than once during a period.
The cleverest students’ results were better than the existing algorithm.
The researchers then analyzed how the best of the students approached the problem and found that in most cases, it was through a known high-level strategy called linear temporal logic. Looking at something being true until something else makes it not true is part of it.
The researchers then encoded the strategies into a machine readable form.
That, along with other human cognition and instinct analysis, is part of what’s going to be behind MI’s leap forward. “Cognitive Agents,” Deloitte calls it.
The MIT students were able to “improve the performance of [existing] competition-winning planning algorithms by 10 to 15 percent on a challenging set of problems,” MIT News says, of the logic it plans to copy.
This article is published as part of the IDG Contributor Network. Want to Join?