Source – cio.com.au
There’s a significant level of hype around machine learning – generated by tech vendors, the media, and even CIOs themselves. But putting the hoopla aside, machine learning technologies actually have real substance and as a CIO, it’s worthwhile getting on top of the trends in this area.
So what’s all the fuss about?
Machine learning is here and we use it every day without considering that it’s actually what is powering many of the services that we use. A great example is Google Maps, which most people would be using daily or weekly.
Initially, Google used machine learning to review images and to protect the privacy of users. However, the company soon realised the technology could be used to automatically provide up-to-date information to Google Maps – and read street numbers and names, and even business names from images.
Some of us also use machine language for speech recognition – like the technology used in Apple’s Siri personal assistant.
In the financial services sector, several innovative banks have applied technologies as (nuance) to integrate voice into their digital channels. This is not new technology but note that Google, Hitachi, Samsung and others are all looking at artificial intelligence (AI) and voice integration. In essence, what all the fuss is about is that we now have a more advanced ability to apply machine learning to automate what were previously quite manual tasks.
Machine learning for the IT team
We have all heard about IBM applying Watson to learn about treating cancer, but I’m also aware of real examples of this technology being applied to deep learning on the operation of production IT systems.
By looking for trends and reviewing all those logs and alerts, then we can see that machine learning has a significant leg-up on human beings ability to absorb all the details and synthesize this same data into an insight.
The opportunity for the CIO is to get in front of this change and start to evaluate how this could be applied in your own organisation. One of the key things that you can bring to the discussion is that the understanding that machine learning is just data analysis, with along with neural networks and AI, has been around for many years.
So what’s different? Quite simply, it’s the availability of cheap cloud storage.
Machine learning is not big data but relies on having large amounts of information to be processed. A great example is the use of machine learning technologies to evaluate an existing dataset of insurance claims and look for patterns of fraud. By just reducing this current state by 1 per cent is a big deal and would provide immediate payback for any insurer.
So what should you do?
While the average CIO is never looking to add another project to their already full portfolio, this is one that doing a proof-of-concept will really pay dividends. For me the worst scenario is that you simply ignore this request and this happens without IT involvement.
In reality, to perform such analysis requires new competencies and mindset that your existing team most likely does not possess. This is a great learning opportunity and in most enterprises this can’t be successfully achieved without buy-in from the CIO.
There are many examples of machine learning being applied to chat bots for instance. Brands like Mastercard, Starbucks, CNN, Pizza Hut and Dominos already are using such technology.
A few IT helpdesk examples are starting to also emerge and I’m sure if you analyse your statistics, it’s likely that there’s an opportunity to reduce the number and the impact of these incidents.
A really simple exercise is to look at your own helpdesk and ask, ‘can we analyse the dataset to reduce the number of support issues?’
For instance, the newer players in this segment such as ServiceNow are already incorporating a machine learning engine to predict outages, automate routing and workflow, and benchmark operational performance.
Now where’s the downside in reducing your own support burden?