Artificial Intelligence for Game Testing: A Friend or a Foe?

by Kate Libbie | July 2, 2020 11:54 am

The game industry is developing at a rapid pace, and game testing, in turn, becomes a more puzzled and extensive one. Meantime, the advent of technologies powered with Artificial Intelligence and machine learning algorithms promises to automate a whole slew of test cases and make game testing smoother. But how does it look like? What is the current level of AI development we have today in the game industry? Let’s explore it. 

How Can Artificial Intelligence Help Game Testers?

Just a couple of decades ago, many “thought calmly” that they would never succumb to the soulless automation processes on the part of technology and AI, in particular. But now, AI-powered solutions, just like the legendary Terminator, are invisible: they never know fatigue, they don’t forget anything, and act only as they are programmed to act. So what can it bring to game testing? 

Let’s face reality, games have always been the toughest pieces of software to test. Compared to websites or mobile applications, games require an almost infinite number of states, incredibly custom interaction models, frequent updates, and make so much money the team cannot afford the risk of running a wrong version. All this means only one: this field requires applying new approaches. 

[1]

Over the past few years, AI has succeeded in performing human-like tasks. Playing games was not an exception. AI can play not just chess but also Go, and even League of Legends. These projects leverage a complex combination of deep neural networks, reinforcement learning, classifiers, and put tasks earlier done by humans to action. But what about testing?

When it comes to game testing, the methodology of using AI doesn’t lag behind. In case when there are myriads of problems connected with manual testing, ‘machine-mind’ can check everything much faster than QA engineers do. Thanks to allowing to select needed parameters, the cutting-edge technology executes thousands of test cases regularly in just 10-15 minutes.

Just think about it. If we develop the right AI with pre-trained experience, it can quickly predict precisely where the more significant amount of bottlenecks are hidden. And if AI-solutions get the opportunity to test a lot of other games, these systems will gain invaluable experience, which can be useful to make the right decisions on the deployment of verification techniques. But this is not all good things about AI for game testing. 

Apart from the advantages described above, AI notices every little thing that could potentially be removed from the structure of a game product. In cases when manual automation skips significant changes in the latest build, AI automatically clicks on the button and draws the user’s attention to the picture that disappeared from the application. 

How Does AI Know What Data to Use for Game Testing?

90% of the flow of information inside the game consists of template data that can be effectively classified and structured. Field names, email addresses, phone numbers, search queries for a specific profile, media data, in-app purchase issues, and more – a lot of applications operate on the same data. Even a small set of input data will be enough to build a network of productive testing based on the achievements of AI. At the very least, all this is enough for you to have a reliable and trusted assistant at the development and testing stage.

How Does AI Understand When Game Is Operating Well?

For testing games, QA team[2] often may have a set of some tests, for example, 100 automated test scripts, they run well, but their functionality is tailored solely for checking only a small space or structure. But, does everything work as it should or not? And here is the right time when AI comes to play. AI-based solutions can solve hundred and thousands of test cases to find an error. For this reason, it could be the right path for game testers to understand whether a game functionality corresponds to the desired plan. 

AI-Powered Tools to test Games More Effectively 

Right now, it is possible to use some tools based on AI, which greatly helps testers to perform their functional duties efficiently. Here are some of them:

PROWLER.io[3]

When there is a need to use even the simplest scripts, you still have the risk that as a result of any change, the script may break. But if you take advantage of the PROWLER.io tool and connect it to continuous integration, you will have an excellent opportunity to identify the consequences of changes made to the game. Besides, by doing so, you can also replace existing test scripts and complement user testing.

Sauce labs[4]

One of the first programs to run tests in the cloud. The service launches up to 1 million automated tests every day. Based on the experience of machine learning, the developers of Sauce labs are working to create powerful tools for analyzing the quality of products.

Test.ai[5]

This tool operates on the basis of AI and Selenium, a popular Automated framework in the circles of testers. Using it, you can perform test cases without the necessity to code understand the mass of locators. Tests are created in a simple format. 

Applitools[6]

Last but not the least tool that is good for checking user interface. Using some of the best practices of machine learning and AI, testers can easily find inaccuracies in the user interface. The application allows you to quickly adjust the test format to the necessary display forms (adaptive view, desktop resolution). If your product uses animation, you can program the utility to find it.

Finally, Insights

The usage of the above-mentioned tools can significantly facilitate your game testing process. Nevertheless, there is no sense to expect wonders from ready-to-go solutions existed on the market. They will do only a small part of a job. In order to use the AI-powered solution that will learn from its previous experience, you have to develop it by yourself. More complex tasks can be solved only with individually build Artificial Intelligence for your needs.

…………………………

If your products are requiring testing, we are ready to hedge your team and ensure the high quality. We are participating in the life of the product, and do our work with a passion for your success. Let’s make great products together.

Visit our blog[7] to read more about QA & testing. Feel free to contact us[8] for collaboration.

Related Posts:

• Software Testing for Hyper-Casual Games: Simple Doesn’t Mean Easy[9]
• Machine Learning Use Cases: what to expect[10]
• Will Machine Translation Steal Interpreters’ Job?[11]

Source URL: https://blog.qatestlab.com/2020/07/02/ai-game-testing/

---