subscribe: Daily Newsletter
search the site
Artificial intelligence for data analytics
Kathy Gibson reports from MyWorld of Tomorrow – Artificial intelligence (AI) could finally be coming into its own as a means of analysing the vast quantities of data being produced by the Internet of Things (IoT).
Carl Wocke, founder of Merlynn, points out that one of the biggest challenges with the IoT is to build appropriate logic into these systems.
Traditional data mining is about building insights using historical data. “But there is a challenge around the scale and width of the data,” says Wocke. “Data mining is a massive challenge and you need an IBM Watson to cope with that. And even then hardware cannot cope with all the data.”
The other option – and usually the best solution, he says – is to ask a human.
“This is what we do. We believe the world will be a better place if we can get computers that think like key individuals.”
Wocke set up Merlynn to attempt to work out how people think, and to clone the way they make decisions. It’s what he calls the “grandfather dream”, where he imagines a world where you could talk to your grandparents even after they are gone.
“Knowledge and insight disappear when someone dies,” he says. “The concept was to build a framework that can house this intelligence. Merlynn is in the people-cloning business, with specific emphasis on decision cloning. We create virtual versions of your decisioning capabilities.”
Wocke points out that the IoT is about getting all the various data elements to talk to each other, and share what they are experiencing. Typically these data points would draw conclusions from various external factors.
“What I would like to see is that the intelligence going into virtual experts,” Wocke says. “We would rather bring in the expertise of real people.”
Examples of systems that already have human inspired thinking include a war room, where there is a layer of expertise making mission-ecisions all the time. In humans, the more experts make decision, the more they occur on a subconscious level.
“In a war room context you have key individuals making decisions. But they get stressed then go home. We now have the ability to staff the war room with experts who can assess millions of transactions per second without ever getting tired.”
In the call centre space, a local company has built a system based on their top call centre agent. In fact, says Wocke, the virtual agent is more efficient.
Human capital management using artificial intelligence can help with staff retention by scanning all the profiles of all the staff members of an organisation every day.
A forex risk agent can mimic three traders, which helps to obviate risk.
“We have the ability to create a virtual version of you that looks at how you operate and comments when you act in an atypical fashion,” Wocke says.
When the state of Utah identified a problem with recidivism, Wocke created four virtual sheriffs that can scan the profiles of all released offenders; each scans for different characteristics to predict behaviour and one of them recommends which action to take.
Stock control is a problem in many organisations that can be solved by virtual agents.
And even virtual doctors are making an appearance, helping to make key decisions that could save lives.
“Imagine an interconnected system acknowledging multiple tiers of inputs and creating insights as if domain specialists were commenting in realtime.
“It’s about making millisecond decisions that require a range of inputs. It allows you to assess vast amounts of data and assess them through the eyes of experts.”
Merlynn’s technology is Tacit Object Modeller (TOM). Tacit knowledge could be equated to “gut instinct”, says Wocke. “TOM uses the latest advances in machine learning to learn from your top experts, not from big data. It looks at how they make decisions – what they consider and what they ignore; what the sanity checks are; and what the decision rules are.”
TOM has been field tested for over five years in a range of industries including healthcare, insurance, banking and recruitment.
The system can be made infinitely scalable as a decision support tool for field workers, as a virtual advisor, and as a training application for less experienced workers and to domain specialist consistency.
Within the next few months, Merlynn plans to launch www.uptotom.com, where individuals can go online and create virtual clones of themselves.