AI still needs teams of human intelligence
Accelerating innovation with hackathons

How can you accelerate innovation to help large companies to find solutions to large and complex problems, quickly and with radical new approaches? Dootrix’s MD, Rob Borley, explains how hackathons can provide the answer.
“It’s completely mind-blowing how strangers can form a team and deliver amazing work in such a short space of time. It really reaffirms my faith in people and what they can do together.”
Individuals, teams and departments could and often do find new ways of doing things better, but progress is usually a slow and incremental, so how do you initiate and speed up radical change that not only gives you competitive advantage, but also saves money?
One of the most effective ways is to give your best minds the space and freedom to operate in an environment where ‘the rules are: there ain’t no rules.’ Strip away the constraints of the office and accepted corporate ways of doing things, and say “never mind how we normally do it, how would you do it?”
The freedom to truly start with “what if…?” is one of the real strengths of a well-run and tightly focussed hackathon. You bring together teams of people from disciplines and departments who wouldn’t normally work together; set them a realistic objective with a clear idea of the desired outcome; give them an energetic and supportive atmosphere; plenty of fuel (good food), and create an environment that allows the discovery and exploration of new ideas, as quickly as possible.
The Dootrix team has run and participated in several successful hackathons, so we were delighted to join Clancy Docwra’s team for Northumbrian Water’s Innovation Festival, to try and solve the problem of water leakage. Leakage costs the water industry millions of pounds each year in emergency repairs and regulatory fines, but the vast and often ancient network of water-mains is complex, and the operational data is doubly so.
The Holy Grail of the water industry is to be able to predict where and when a leak will occur and to predict how much will it cost to fix before a small problem becomes a very big one.
How do you predict the future?
The answer is to interrogate the past, using a combination of Artificial Intelligence (machine learning) and the collaborative might of human intelligence and sharp industry insight.
Using Microsoft’s Azure Machine Learning Studio (and with plenty of help from their incredibly helpful and knowledgeable ‘evangelists’), our team of designers, developers and analysts set off to find patterns in the historical industry data that might give clues to the future.
To get powerful Machine Learning systems to work, first skilled humans have to give it the right data to interrogate, so it can teach itself to find the desired patterns. Where human brain-power comes into it’s own is in setting smart, real-world, parameters for the AI machine to learn from. If you want AI to drive from point A to point B it first needs to know if it’s going to be driving a car or a train.
In the water industry, bursts and leaks are very different things — slow seepage versus a sudden gush — and fixing a dribble is cheaper and easier than fixing an unexpected raging torrent. But it’s much easier to collect relatively accurate data about occasional bursts than it is to record numerous small leaks, so the burst records were a solid place to work from in the brief 48 hours of the hackathon.
The age and construction of water mains varies incredibly, from Edwardian cast iron to 21st century PVC, and the pressure, flow rate, weather, surrounding soil and overhead traffic all will have an obvious effect on the mains’ integrity, but immediately that adds a multitude of variables about the past, not the future.
What our team of man and machine came up with was a prototype that showed it was possible, with further evolution, to use historical data to give a likely longevity of any given water main in the future, and the probability of where bursts would likely to occur. Accurately defining the precise where or when is the next job, but the team’s award-winning idea showed that by harnessing AI to the industry’s ‘big data’ with a “what if?” approach to innovative thinking, can deliver huge cost-savings in real world situations.
Charlie Allen, our Delivery Director has run two successful hackathons for Manchester City Football Club so we sent him to the NWG hackathon as a participant, and he’s in no doubt of their value.
“Hackathons are a great vehicle to validate and prototype ideas rapidly, which can then be turned into real products and services. Collaborating with the guys at Clancy & NWG gave us a level of industry insight and domain expertise that was essential for getting straight to the heart of the problem, and from there, finding a simple solution that with further evolution could be deployed quickly in the ‘real’ world.
“We all like challenges, new experiences and to be taken out of our comfort zone. It’s good to keep moving, to move quick and well. Momentum and continuous learning are not only at the heart of innovation, but also are integral to success when designing and delivering a hackathon.”
We asked Charlie to write down his recipe for a successful hackathon, which you can read here, but if your organisation simply needs to stimulate and accelerate innovation, or just wants to understand how you can use AI to crack your toughest problems, please do get in touch with us.
