I recently came across a Brookings Institute research article on What jobs are affected by AI? Better-paid, better-educated workers face the most exposure. The findings really intrigued me since they appear counter-intuitive to conventional wisdom that AI will have the most impact on blue-collar workers as a result of automation. This research was performed using a novel approach of looking at patent filings on AI and correlating key phrases to job descriptions to determine which jobs will likely get automated and hence high likelihood of being impacted.
I work for a company (sofy.ai) that has multiple patents in automated testing powered by machine learning. Hence, his research provoked my interest and prompted me to write this blog.
The stigma of AI & Machines Replacing Humans
We are a test automation company and I have family members who are employed as software testers with whom I have had discussions on whether companies like us will replace their jobs.
When I have discussions with some system integrators (SIs) about adoption of automated testing, I sense a level of apprehension as such technologies are seen as replacing human testers and jeopardizing their primary business model for revenue generation which is through placing people as billable resources with customers. While others are forward-thinking and see such innovation as an opportunity to compliment and perhaps deliver services that are not possible with manual testing alone. After having these discussions, the question for me is “will technologies like Sofy replace manual testing as we know it?”
Technologies like Sofy augment manual testing by performing tasks that are impossible to execute manually or meet the rigor of testing needed in the era of digital transformation which cannot be fulfilled through old paradigms. The problems are particularly demanding in the mobile testing space. In addition, AI enables new scenarios which opens new possibilities.
I recently started using a voice transcription app called Otter. Did this replace what would have been a job of an admin in the yesteryears? Perhaps in some ways, yes, but that is not the reason I use Otter. It allows me to communicate differently and more effectively with my team and to do so at a different pace. As I write this blog, I am traveling, jetlagged and awake at 4AM and can use this technology without requiring the presence of an admin. So back to my question, how then is Sofy augmenting manual functional and UI.
3 Ways AI augments Manual Software Testing
In the mobile and webspace in which we operate, the pace of release is measured in weeks. As a result, manual functional and UI testing is simply not possible without some level of automation.
Most companies undertake exhaustive testing only for major releases since it is simply impossible to do exhaustive testing for every major and minor release. The downside of this approach is that it is hard to predict which issues will resurface in regression without comprehensive testing and compromise the quality of software. Compounding the problem is the need to validate the app across plethora devices and OS. As a result, companies are forced to choose between testing for quality or speed of release. With automation, these issues can be addressed as an automated software does not sleep, suffer from fatigue or make mistakes and costs a lot less.
Deterministic and non-deterministic testing
AI testing can perform both deterministic and non-deterministic testing. It is impossible to do the latter with manual testing. You just cannot simply tell a human tester to perform a non-deterministic test and do so every day or many times a day due to what we refer to as “muscle memory”. One of our customers chose Sofy because of this capability as Sofy discovered crashes in their software that humans could never detect.
Testing for quality and performance
Testing is more than finding software bugs; it is also about finding flaws in UX design, lack of accessibility and fine-tuning performance at a page or screen level. These types of issues are typically never the preview of manual testing. With AI and automation, it is now possible to validate the app for these types of issues and ultimately provide the user a great experience.
By leveraging AI, it is possible for product managers, developers, QA testers to have a real stake in developing tests once with right domain expertise which can then executed as often as is needed through the life of the app. Finally, as the trend of combining development with testing becomes more pervasive within organizations, developers don’t want to perform push button or manual testing of UI. One of our customers faced this exact situation which prompted the search for an automation tool.
Regardless, of what automation can do, there are some aspects of human testing that automation can never address. Human testers are not going away, but the industry will have to consider automated testing as another tool that helps them deliver software to their customers.
This is precisely the mission of Sofy. Check us out at https://sofy.ai.