Augmented reality, machine learning and the cleric

Autonomy CEO explains how the future of computing stems from the 18th century...
Written by Nick Heath, Contributor

Autonomy CEO explains how the future of computing stems from the 18th century...

When Presbyterian minister and mathematician Thomas Bayes put quill to paper in the 18th century, little could he know that one day his equations would help meld the virtual and physical world.

But more than 200 years after Bayes' death, Mike Lynch, CEO of Europe's second largest software company, Autonomy, is arguing that machine-learning software built on Bayes' theorem on probabilistic relationships will underpin the next major shift in computing - the move to augmented reality (AR).

"One of the biggest areas [of computing] is going to be in the area of augmented reality - it takes the online world and slaps it right in the middle of the real world," Lynch said, speaking to silicon.com at the recent Intellect Annual Regent Conference 2011.

Today, augmented reality apps run on smartphones, layering digital information over video of the real world taken by the phone camera in real time. But in future, the apps could lay digital information directly over everything we see, using screens built into glasses or contact lenses.

Augmented reality app for Apple iPhone by Museum of London

An augmented reality app for the Apple iPhone from the Museum of London lays historic images over London landmarks
Photo: Museum of London

Smartphone apps already exist that do things such as lay historic photos over images of London landmarks, but Lynch said AR will eventually permeate our lives - putting the digital world at the heart of everyday interactions.

"Perhaps a printed poster on the wall becomes animated, and you can click on it and buy the DVD - suddenly what was a simple ad becomes a way you can buy something," he said.

"Or you're walking around London and you hold up your phone to a statue of Eros and it tells you the history of it.

"Or you meet someone on the street, hold up your phone and it tells you about what they're interested in, and maybe in the virtual world they also have a parrot sitting on their shoulder.

"It's a completely different way of interacting with vast amounts of information in situ and in context."

Machine learning

Many AR apps available today rely on GPS and digital compasses to work out what the phone is pointing at and what information to display, but future AR apps will...

...increasingly need to understand what the user is looking at and what digital information they want to see, a process that will require machine learning.

"Everything we are talking about comes down to the ability of the computer to understand what something means," Lynch said.

"It's the mathematics of Thomas Bayes that allows computers to learn what things mean.

"It's a self-learning system - so basically by reading the newspapers [a machine] learns all about our world. For example, a computer could learn that 'Becks' is David Beckham, and that he's married to 'Posh', that he's very good at football and a bit of a fashion icon."

Drowning in data

Lynch's vision of the near future is a nice fit for Autonomy and its specialism in machine-learning and pattern-recognition software that can analyse unstructured data - information that has not been labelled and linked to other information inside a database, where it can be read and understood by a machine.

Since Autonomy was founded in Cambridge in 1996, the company has been helping businesses tackle the tide of unstructured information that flows into a modern business.

Today, Autonomy has a market capitalisation of $7bn and a customer list that includes more than 20,000 major organisations worldwide - including BAE Systems, the BBC, GlaxoSmithKline, Nasa and the Houses of Parliament.

The amount of unstructured information - whether it is text in an email or an audio recording of a phone call - is growing so quickly that Lynch believes organisations will soon have no choice but to task machines with analytical work that previously would have been the preserve of humans.

"Some 85 per cent of what you deal with at work is unstructured information," Lynch said.

"You can replace people in lots of tasks where people are looking at unstructured information - for example, reading an email and routing it to someone else, looking at security camera footage or going through documents to find which are relevant to a law suit.

"If you can get a computer to do [those tasks] then that's a phenomenal saving, and it frees up the human to do something more interesting.

"It's going to have to be that way because the amount of unstructured information is growing at 67 per cent [each year] - so if you are going to use people you better get breeding."

Perhaps in a nod to the rise of AR, Lynch said the most valuable lesson he had learnt since starting Autonomy was that the tech industry is built on shifting sands.

"We always think everything is set in stone and this is how it is. For example, Microsoft dominates the industry. The one thing you learn is nothing is set in stone, all the stones are moving, there's incredible opportunity all over the place and the fat lady has not sung," he said.

But even as technology accelerates the pace of change, and the digital world becomes intertwined with the physical, Lynch takes comfort that, AR future or not, some things will never change.

"I live in Suffolk and the nice thing about Suffolk is that the conversation down the pub is the same as it has been for the last 500 years, which is 'How do you get rid of moles?'," he said.

And as confident as he is in taming the world's information, Lynch admits this is one challenge that has got him beat, conceding: "It probably will always be an unsolvable problem."

    If you like this story then you may also be interested in silicon.com's other articles on augmented reality and artificial intelligence:

  • Photos: Augmented reality apps for iPhone and Android
  • Cheat Sheet: Augmented reality
  • Cheat Sheet: Artificial Intelligence
  • Artificial intelligence: Can AI crack the conundrum of consciousness
  • Artificial intelligence: How advanced will AI be in 2100

Editorial standards