Generative Artificial Intelligence (AI) has emerged as a powerful tool for automating software testing. For the first time in history we have algorithms ability to replicate human logic, Generative AI has the potential to save engineering and product teams a significant amount of time in creating user stories every time there is a product update. In this article, we'll take a closer look at how Generative AI can replicate human logic for software testing, provide examples of its effectiveness, and highlight the time-saving benefits for engineering and product teams.

At its core, Generative AI is a machine learning technique that enables computers to generate new data and insights based on patterns and relationships found in existing data. In the context of software testing, Generative AI can be used to automatically create and execute test cases that mimic the logic used by human testers. This means that software can be tested faster and more thoroughly than ever before, with fewer errors and greater accuracy.

We spoke to 11 QA experts from Big Tech firms and this is what they said. They spoke anonymously to speak freely.

1. Google

One of the key advantages of Generative AI is its ability to learn and adapt to new situations. For example, a Generative AI system could be trained to recognize and respond to new types of bugs or errors as they arise, without the need for human intervention. This not only saves time, but it also enables engineering and product teams to stay ahead of the curve when it comes to software quality and reliability. Google teams have been using their own LLM models for generating User Stories for many Google products esp. in cumbersome multi-device scenarios, where they can go wrong pretty quickly.

The most promising models have are being used to test software with high amount of dynamization like traffic prediction UX on Google Maps.

2. Facebook

Another advantage of Generative AI is its ability to identify patterns and relationships that might otherwise go unnoticed by human testers. By analysing large amounts of data and testing scenarios, Generative AI can identify hidden bugs and issues that would be difficult or impossible for human testers to detect with naked AI. This is especially true for testing User Experience and iconographic or colour preferences. Facebook teams are using automation more and more on social products, comment section management, graphics for discovery of safety features etc.

3. Amazon

Generative AI can also be used to automate repetitive and time-consuming tasks that are traditionally performed by human testers. For example, a Generative AI system could be used to automatically create and execute test cases for multiple versions of a software product, or to test software across a variety of different devices and platforms. This can free up valuable time for engineering and product teams, enabling them to focus on more strategic and high-value tasks. Amazon teams have used home-grown models to test e-commerce listing page dynamization. However it is not very clear how much Amazon has invested in own LLMs compared to other players and the company may look to acquire available technology soon to catch up with peers.

4. Microsoft

Microsoft has at present clear head start on use of Generative AI with the acquisition of Open AI. Microsoft has already integrated ChatGPT into Bing Conversational threads and is using that internally to automate faster product release cycles. One of the most exciting aspects of Generative AI is its potential to help engineering and product teams keep pace with the rapid pace of software development. This is a critical yet overlooked factor in software development. As software products become more complex and sophisticated, the amount of testing required to ensure quality and reliability can quickly become overwhelming. Generative AI can help to automate much of this testing, enabling engineering and product teams to focus on innovation and new feature development. Using LLM models, Microsoft team could release the Bing integration to ChatGPT literally in days vs. months or years.