A Merger of Equals: Human-Machine Teaming My car broke down last night.
Luckily it was on a short trip to the supermarket rather than a longer journey but still, it was hardly an ideal scenario when out late at night.
It got me thinking though. Grateful as I was for the assistance from a local repair service, I couldn’t help but wish that I’d had some forewarning of imminent problems, rather than have the car grind to a halt seemingly out of nowhere. Thankfully, this type of scenario might soon be a thing of a past.
Engines with a full complement of sensors – unlike that on my own car – produce a huge amount of data at a pretty impressive rate.  Although it would be next to impossible for a human to accurately analyse all that data in time, algorithms can spot patterns that are indicative of impending faults or just plain unusual, and let the human know. The driver, who is the expert in the engine and understands the context, can then decide what to do about it based on factors that the computer couldn’t possibly know about – traffic, time of day, weather, nearest garage and so on. 
This is an example of what I call ‘Human-Machine Teaming’.
It is already standard in some specialised industries: fighter jets, airliners and helicopters, now it’s starting to move into helping normal people go about their everyday lives.

Teaming up

Clearly, data can help a person improve their decision-making but often that involves them wading through lines and lines of the stuff. This is cumbersome and very dependent on someone knowing how to do their job inside out – and if they miss something then there are problems. Human-Machine Teaming, by contrast, is all about making the users lives easier by presenting what they need right there and then.
So, how do we do this? 
As a part of Programme Nelson within the Royal Navy – specifically the Data Platform Development Team which I am Technology Lead for - we’re creating a new platform similar to popular mobile application stores, one where the company can operationalise its data in one place and make it accessible to everyone. This allows all parts of the organisation to use the data and get creative with it – you can mix it together, you can combine it, you can do interesting things with it in an operational context.
But in order for the applications an organisation uses to process the data to be trusted and useful, the platform and the data have to be consistent and reliable, scalable and accessible. For this to happen, you need data engineering and data science to come together. 

The importance of engineering excellence

Engineering makes things robust, reliable and usable. Without this foundation, the other pieces of the digital jigsaw can’t perform. This is because data engineering is about concepts. It is about providing a robust, secure, scalable solution to a challenge.
Technologies are always changing all the time but data engineering is about more than the latest wizardry. After all, an architect can design a beautiful building, but it’s an engineer who builds one that lasts for 100 years – and the same applies in the digital world.
If you’re a data scientist like me, you’re looking at new algorithms. If you’re a data engineer, you want to make the concept real. You want to build something useful, practical and reliable that will genuinely help users get their jobs done faster and better. A lot of data scientists can be happy that they have written a new algorithm but is anyone actually going to use it? That’s where data engineers can help – it’s much more about making things a reality for customers, making sure those algorithms are used to provide value.

Eyes on the prize

Currently, we are well on the way to turning the vision for our data platform into impact. Although still a work in progress, it offers a taste of the type of outcome that can occur when humans and machines can combine their strengths – such as predictive maintenance for a car engine.
The pace of technological development is only going to accelerate. It’s up to us, as humans, to ensure this type of result becomes the norm, rather than the exception. Human-Machine Teaming is one way of doing this.
About the author
Hannah Green is a Lead Data Scientist with BAE Systems Applied Intelligence

Hannah Green

Lead Data Scientist, BAE Systems Applied Intelligence