View from the crest of the 2016 data wave
At the close of 2016, some reflections on the world of data analytics, machine learning and AI.
There has been an incredible buzz about data this year. The hype level concerning data analytics and machine learning is increasing. We are seeing very significant investments from Microsoft Ventures, a venture capital fund exclusively focusing on AI startup, to Intel acquiring Nervana for £300m to develop their expertise at deploying machine learning onto the silicon layer itself. With all the rising expectations that data analytics and machine learning will cure cancer, bring world peace and deliver snow at Christmas, we might fear that the hype bubble will burst. However, having been around data and AI since the late 1980s I think we now have real reason to be excited. After all, who last manually focused a camera? The mass personalisation of products and services is becoming ubiquitous, and is often powered by machine learning. The value is really delivered when smart computing systems – be they analytics, machine learning or AI based – are embedded into everyday processes and activities in ways that just work, making them better for humans. It’s worth noting some key themes:
- Servification is ensuring the centrality of data – where people focus on the provision of a service (as opposed to a product) data become essential to driving the business. From those who strive to emulate the success as ‘service only’ – companies from the likes of Uber and AirBnB, through to industrial giants like Hitachi with their ‘train as a service’ for the Great West mainline – they are gathering and exploiting data like never before to deliver the marginal improvements that drive business effectiveness.
- Automation – for many customer facing businesses, there’s a huge move towards automated customer service, driving toward self service and automating as much of the back office processing as possible. Data analytics steers what can and should be provisioned as self service. AI provides the 'smarts' to automate more complex tasks successfully.
- Data as an Asset – from corporate giants to local authorities there is a growing recognition that data isn’t just the by-product of their core operations but could be an asset in its own right. Their financial director manages their money as an asset, their HR director develops their employees as an asset, and their facilities management service looks after their buildings and physical assets. But who curates their data as an asset, ensures they’re getting best value from their data, commercialising it where possible and leveraging it for greatest advantage? Perhaps their yet-to-be-appointed chief data officer?
- Platform as a Service is coming of age – the cloud vendors, such as Microsoft’s Azure with Data Lake Analytics and IBM’s Bluemix with Data Science Experience, are making it possible to develop a rich and diverse compute and storage infrastructure for ‘big data analytics’, that will scale in volume, processing complexity and data variety.
- Mainstreaming open source for analytics – increasingly, analytics teams in corporates are switching away from traditional statistical tools such as SAS and SPSS, toward open source tools like R and python. Along with this is a shift in skills away from button pressing in an application toward crafting code. However, insightful interpretation and application of analytical output remain the vital ingredients to deliver business value.
- Information governance and data ethics are on the rise – whilst the confusion abounds in a post-Brexit-vote Britain, there is a growing realisation that we need to grapple seriously with data ethics. We saw the government agreeing with the Science and Technology Select Committee’s recommendation to establish a Council of Data Science Ethics and, notwithstanding any Brexit negotiations, we all need to be ready for GDPR.
- Growth in multi-skilled teams – the most interesting breakthroughs are likely to come from collaborative endeavour. Concepts such as an Innovation Lab that draws together disparate skills such as creative designers, behavioural scientists as well as data analytics, machine learning and insight experts to focus together on a problem are a very potent. Though we still need to learn a couple of things. Firstly, how to enable diverse temperaments to best work together and not succumb to the ‘tyranny of the extrovert’ whereby the loudest and most confident always dominates. Secondly, how to bring business nous into the environment, to steer innovation towards commercially viable approaches, in such a way that pragmatism doesn’t stifle creativity.
Roll on 2017, with all the data it will bring – new streams of machine generated data from IoT and user generated data from sensors. There’s a growing shift in consumer expectations of how smart computer-based services ought to be … “what do you mean you can’t predict what I want for Christmas next year?”