Artificial intelligence, once the preserve of science fiction, has gone mainstream. Whether it’s Elon Musk feuding with Mark Zuckerburg, or Amazon Alexa fighting crime, scarcely a week passes without the media scaremongering about AI’s effects or speculating on its future.
Venture, corporate and seed investors aren’t immune to the hype. There’s been something of a gold rush to fund AI and machine learning solutions (already to the tune of $3.6 billion), and you don’t have to look far to find examples of industries likely to get swept away by the nascent technology.
Artificial Intelligence is predicted to eliminate 6% of US jobs by the year 2021. The question now for many commentators is which industries face the chopping block, and how soon we should start worrying.
It’s not just low level or menial jobs threatened by AI. Start-ups such as Oxford University spin-out DiffBlue have made headlines for using AI to revolutionise the way software developers work. DiffBlue recently raised £17 million for its AI solution that automates coding tasks considered an inefficient use of a human developer’s time.
In the coding community, attention has turned to AI’s ability to disrupt the ‘testing’ industry entirely. Are the days of manual testing soon coming to an end?
We think it’s unlikely. Although AI is already revolutionising the way software developers having to test complex chunks of code (and rightly so), it has a long way to go before it can accurately replicate aspects of user-based testing. The number of different scenarios associated with a human interacting with an app or website are close to limitless. AI doesn’t currently have the capacity to successfully tackle the multifaceted and unpredictable variables associated with how, where and when humans use a particular app or online service.
This is particularly true on mobile. An initially poor mobile experience will discourage 85% of customers from further use, studies have shown. Perfect user experience is more important than ever to consumers, and human-based testing remains the most assured route to securing it.
The human shopping experience
Consider using a taxi app to order a car to arrive at your home on a rainy day, only to find that your driver has driven to the café 500m down the road instead. Noticing this, you walk to your car, getting soaked in the process. The app’s system registers a successful pickup. You, on the other hand, are far from happy.
Or imagine you go to an electrical store, to pick up a new laptop pre-ordered on the website. When you arrive at the store it takes ages to find the right one, and there aren’t enough staff members on hand to assist you. At this point, some people would have given up and gone to a competitor.
The detection of user frustration and its contributing factors is difficult unless you have resources on the ground to test for real-life scenarios like this. Also, companies have traditionally focused on small sample sizes of testers due to financial constraints.
The crowd testing solution
Crowd testing puts digital properties (such as mobile apps, websites, IoT and connected devices) into the hands of people that are representative of your customer demographic. With crowd testing, you can expand the sample sizes of testers while simultaneously focusing on the most relevant consumer demographics for the product. For example, if you require men aged between 18 – 34 years old in the South West of England, a crowd testing company can use its community of testers to get your product tested by the core demographic.
This pool of testers is then further refined as it can be split into two groups: the vetted quality assurance (QA) professionals and the ordinary user with no QA background. The former allows experts to effectively identify any bugs or problems that might have been missed by your internal team, while the latter will allow you to explore how intuitive your product is. Being able to accurately test how your website or mobile app will function ahead of its release in a specific market gives you an invaluable opportunity to ‘get it right’ before your customers start to engage.
So is AI going to replace humans?
AI can identify errors and bugs in software with unparalleled precision. AI doesn’t rely on food, water or sleep and, unlike a human tester, can undertake repetitive tasks 24/7. As the technology continues to develop it will increasingly become an integral part of the troubleshooting process.
But this is not to say human testing will ever become obsolete, and therefore businesses must not treat it as a panacea for all potential problems.
While AI might be ideally suited to spot errors in code, it still lacks the sophistication required to adequately reproduce a human’s user experience. Crowdtesting platforms allow real users to engage with digital products under real world conditions in a way that artificial intelligence can’t.
By using AI, companies have the best chance to perfect their code and UX, removing any potentially harmful bugs. But in an increasingly competitive digital market where first impressions can make or break a product, companies must take a holistic approach to software testing, and I think we’re going to see more and more companies using AI in conjunction with human-based testing. What’s most clear is that first impressions are everything for the consumer, and no company can afford to skip on the user experience.