In recent years, we’ve become used to the fact that retail PCs are cheaper, more functional, lighter and better looking than the ones most organisations provide for work.
Many of us find it more productive to work at home or in a coffee shop, or anywhere there is Wi-Fi. Consumer email, instant messaging, file sharing and other free services are often demonstrably more capable and easier to use than the services that most large organisations provide.
As long ago as April 2004, the Leading Edge Forum (LEF) coined the term consumerisation, and published a report on The Consumerization of Information Technology. The paper was our response to the intriguing developments popping up all around us.
Of course, since then consumerisation has flourished. The tipping point came in 2007 with the arrival of the iPhone, which did three main things. It established consumer mobility as the new IT industry centre of gravity; it made the Blackberry – then the overwhelming enterprise standard – obsolete; and it eliminated the boundaries between work and personal computing. Despite its initial scepticism, it was enterprise IT that had to adjust.
Today, the giants of our industry – Google, Apple, Facebook, Amazon, Microsoft, Samsung, Alibaba – are all rooted in various consumer marketplaces, and consumerisation has become the conventional industry wisdom.
But this doesn’t mean consumerisation has matured, or even that it is fully understood. Consumerisation is still gaining momentum, and is now expanding into areas that few of us imagined in 2004.
The most obvious of these are wearable technologies and the burgeoning “quantified self” movement, but the most important new area is not sufficiently recognised – the consumerisation of machine intelligence (MI), as depicted in the diagram below.
From top-down to bottom-up intelligence
We prefer the term machine intelligence to artificial intelligence (AI) because it is more accurate. Developing advanced MI capabilities has historically been a top-down process.
Many large commercial and government organisations have built proprietary expert systems based on their internal data, software and know-how. These systems play an essential role in all manner of logistical, investment, pricing, command/control and other mission-critical applications. This important work continues to progress.
But just as we saw with PCs, networks, online services and smartphones, consumerisation is now transforming MI into a bottom-up dynamic. This time, the driving force is not technology, but big data.
Many of the most important machine intelligence initiatives today – such as language translation and image, facial, activity and emotion recognition – are based on predictive analytics that get more accurate as the data sets get richer, and consumer markets are where the biggest and best data resides.
An excellent example of this is the image and activity recognition project, ImageNet, which is based on more than a billion downloaded internet images and the work of 50,000 people – using Amazon’s Mechanical Turk to do the necessary cleaning, sorting and labelling.
There is an excellent TED talk by Stanford University’s Fei-Fei Li on this effort. But while ImageNet required significant new human labour, Facebook has a huge head start in facial recognition because it already knows our names and faces. Similarly, Google is a leader in machine translation largely because it has aggregated the best set of multilingual documents.
Deep learning surprise
The speed of these advances has stunned many AI/MI professionals in what was once a relatively slow-moving field.
The latest deep learning surprise was the four-games-to-one triumph of the AlphaGo program developed by DeepMind – now part of Google – over Lee Sedol, arguably the world’s best human Go player. The number of possible moves in Go is orders of magnitude greater than in chess, and few MI experts thought that machines would defeat the top human players so soon.
The February 2016 announcement of a $5m XPrize to be awarded in 2020 for the best TED talk by a machine is further evidence of the MI field’s growing confidence and ambition. Of course, the recent Microsoft Tay fiasco shows that machine intelligence still has a long way to go.
In short, machine intelligence innovation is increasingly dependent on the exabytes of data created by billions of online consumers, and this powerful bottom-up dynamic may well determine the business models of the future.
From an enterprise IT perspective, change is once again coming from the outside in, and IT leaders need to get past the inevitable scepticism and embrace the exciting MI innovations now under way.
While concerns about machine dependency and even human obsolescence are growing and can’t be dismissed, the machine intelligence express will continue to steamroll forward, and – as with previous consumerisation developments – enterprise IT is advised to get on board.
Read more about artificial intelligence
- Socially aware general-purpose artificial intelligence in the form of a dog could be the ideal form factor to take over the world.
- Financial services firm MasterCard is deploying AI software created by one of the IT startups it supports in a global programme.
- Deep learning is generating buzz in the AI community. Microsoft’s Eric Horvitz and Facebook’s Yann LeCun explain why this type of machine learning is so thrilling.
Read more on IT innovation, research and development
MI5 accused of withholding surveillance compliance failures from cabinet minister
MI6 apologises after attempt to interfere with intelligence court
MI5 slammed by watchdog for failing to delete intercepted phone and internet data
MI5 failed to disclose failings in handling intercepted data, court hears