Before we start using large language models (LLMs) to assist our testing, we need to understand how LLMs work, what good testing looks like, and most crucially, how we combine the two to create value. As Marshall McLuhan once stated, “We shape our tools, and thereafter our tools shape us.” This aphorism is also true when working with LLMs. Their behavior can give us the impression that they think and behave like humans, and therefore, we can rely on them to replace our work. However, this kind of thinking can lead us down a dangerous path, and not only slow down our work, but also bias us toward incorrect assumptions about the quality of both our work and our products.
This is why having a balanced mindset toward how LLMs work and how to select specific tasks within the context of testing is vital. We want to use the power of LLMs but also maintain a healthy skepticism about what they return and the influence they have on us during our work. With the correct mindset, the rest will fall into place much easier. So, let’s learn what this mindset looks like.