Machine learning isn't truly new; it's been around for hundreds of years. Even the more modern principles and algorithms behind some current machine learning applications date all the way back to World War II where they were used to do things like find sunken U-boats. The difference is that we now have the computing power and supporting tech to do it in near real-time and on a larger scale.
With this rising tide of technology, machine learning is more available and powerful than ever. It's the culmination of many things - the compute power of your phone, the resolution of your camera, the bandwidth of a cellular network, the availability of machine learning as a service (image identification, voice comprehension, sentiment detection, etc.). So, don't debate whether or not machine learning is available to your business or if it's something you can use. Spend your time figuring out where to use it and how to use it, because if you don't, someone else will.
The place where organizations can separate themselves from the competition by having strong architecture and a service-oriented approach. Most of the work is not the ML itself, but the scaffolding around it. That's why with machine learning, strategy is so important. How can you leverage ML in your business? The first step is figuring out where the low hanging predictions and insights are, and to start building towards that.
This concept underscores one of CapTech's longstanding beliefs: don't do technology for technology's sake. Don't just go into machine learning because everyone else is doing it. Work backward from your business goals before deciding on the best solution to use to reach them.
To prepare, ask yourself questions like:
- Do you want to improve your recommendation systems?
- Do you want to understand what your customers are saying about you on social media?
- Do you want to figure out why something isn't working?
- Do you want to understand your users better than they understand themselves?
Let me give you an example: imagine a production facility where you want to know when parts need to be replaced before they actually fail and halt production. Using machine learning to analyze data from IoT devices, you might be able to recognize when there is a particular temperature change that precipitates a belt breaking two days out.
The same thinking can be applied to many other applications. What if using health monitors like smart watches you could predict when people might have heart attacks before they happened? Or by tapping into employee data, you can find out when people were planning to quit before they tipped their hand?
The key thing to remember about machine learning insights is that they are predictive. While reporting and analytics are still very important to decision making, you can now make predictions of the future at a scale that previously you could only dream of.
Like anything else, machine learning isn't a silver bullet. It won't replace the experts you need to train the machine or the people you'll need to build the supporting data and architecture that make it work right, but it certainly is a head start - so stop waiting and start building.