Original article was published on artificial intelligence
We all know Artificial Intelligence and Machine Learning are transforming business. It’s clear that many companies are rewiring their organisations and creating dedicated teams to capitalise on AI. Although this shift has been happening, up until this point it has been doing so on the fringes, or inconsistently.
2020 will be the year artificial intelligence goes mainstream (if not in spite of, then because of Covid-19, a separate topic altogether). The development platforms, vast processing power and data storage that enable AI are becoming increasingly affordable and more “off the shelf.” Companies are beginning to grasp how to cope with the inherit risks of AI, yet have only just begun to think about how AI can improve every aspect of their value chain.
The age-old business imperative of growing revenue, reducing costs, and enhancing customer experience can be significantly impacted positively with the proper application of AI. This is why virtually every C-Suite is seeing the dramatic need for machine learning applications to be applied to every aspect of business. As a result, according to a recent Gartner study, 9 in 10 CTOs plan to adopt AI in some form in the next three years. Why the sudden interest?
At its simplest, AI is “computers doing things that humans cannot do,” using massive amounts of data. It is the ability to use the growing computational power to help machines and the ecosystem they operate in to learn and react upon consequences. It is not about replacing humans but in enhancing their capabilities – just like any other tool – but this tool is unprecedented in its power and ability to enhance human capacity.
AI’s time has arrived
Simply put, by taking on repetitive tasks and by heuristically learning, AI-driven systems and processes can significantly improve business effectiveness and efficiency. For example, for insurance companies, AI can perform quick, seamless risk insurance underwriting based on massive data, which allows gives the customer quicker decisions and allows employees to allocate more time towards establishing deeper, more meaningful (and sticky) relationships with customers.
For retailers, AI can vastly improve the customer experience by generating relevant offerings that are context specific, more accurately reflective of exhibited consumer tastes, personalised precisely, and offered up in a compelling and seamless way. This invariably results in higher click through rates and ultimate conversion – the nirvana state for any business. It does this by using unimaginable amounts of data inputs and finding trends and connections that no human could predict.
Just ask airlines; other than accidents driven by software errors (like the Boeing 737 Max software issues related to the 2019 crashes), nothing is more damaging to their reputation with customers than flight delays, which not only are a buzz-kill for customers but also drive costs through the roof. The reality of millions of customers each with a unique itinerary and circumstance is simply too complex for any human to deal with. Given the dynamic changes in weather and equipment or staff availability, the only way to solve these situations seamlessly is to apply artificial intelligence.
One such example, is British Airways. To help minimise delays that arise during preparation for departure, BA recently installed cameras at three aircraft stands at Heathrow’s Terminal 5. Delays happen when one of many steps in the aircraft’s preparation procedure encounters an issue that can have negative impact on the whole process.
British Airways is using AI to compare footage of the turnaround process with what is proposed on the schedule. Solving for no delays, if the system notices anything in all the complex systems that need to work together and react without creating bottlenecks that would cause delays, the machine learns over time and can send an alert to the manager in charge through a smart watch, enabling them to act and ensure the flight isn’t held up.
When software code turns into spaghetti
Just as AI is delivering results and enhancing human capacity in basic industries, it is also enabling software engineers to develop more and more complex code – driving increasingly complex systems to be implemented but doing so with less risk – this, in spite of these systems having more code than the human intellect could ever test and validate. In a virtuous cycle, AI allows developers to increase the complexity and functionality of software beyond anything imagined before while at the same time keeping an eye on the quality of that software using AI in testing in ways that could never be done with even the best testing expert on their own.
Code turns to spaghetti when it accumulates over many years, with feature after feature piling on top of and being woven around what’s already there. Eventually the code becomes impossible to follow, let alone to test exhaustively for flaws. Complex software begets complexity in testing. As Nancy Leveson, Professor of Aeronautics at MIT put it: “The problem is that we are attempting to build systems that are beyond our ability to intellectually manage. With 100 million lines of code in cars now, you just cannot anticipate where things might go wrong.”
Airbus’ 30 million lines of code to fly an airplane turned into Tesla’s 100 million lines of code to drive a car in a pretty short period of time. This is where AI applied to testing and quality engineering comes shining through.
AI is the answer to quality assurance in spite of this complexity
To say it more starkly, this is why many companies’ product and service quality is doomed without AI-driven quality engineering, as the biggest role of AI in the life cycle of software development will be played in quality assurance.
In a world where no human being could foresee or predict all of the possible quality issues, machine enhanced quality experts can achieve significant confidence levels in less time using AI in testing. Taking volumes of structured test result data and marrying it with unstructured release notes or agile stories, using machine learning to tell an engineer not only what test to perform or when to perform it but also in what order to conduct the tests can significantly increase confidence in far less time.
AI is here to stay. More and more companies will adopt it into all aspects of their value chain. Such applications will significantly increase the complexity of systems and therefore the business risk of launching this new functionality. But in a way, like a snake swallowing its own tail, AI will allow engineers to keep pushing the boundaries of what software can do as it provides assurance in spite of the complexity. AI, in effect, gives us a lens to look into the increasing complexity, which allows us to confidently test systems that are simply untestable without that aid of Machine Learning. This is why companies that are not using AI in their approach to quality assurance will be at a significant, debilitating disadvantage.
Norm Merritt, CEO, Qualitest