AI for Regression Testing, Generative AI Code Smells and More TGNS99

Does ChatGPT suck for test automation? How can AI help you with regression testing? And what is a real world example of using large language models with rpa? Find out in this episode of the Test Guild News show for the week of October 22nd. So grab your favorite cup of coffee or tea and let's do this. But first, are you looking to take automation testing to the next level? Look no further than Applitools and the Visual AI Validation testing platform. Trust me, it is a wicked game changer. Plus you can you can try it for yourself by creating a free account now by using the special link in the first comment down below and see the difference for yourself. So these news items I'm sharing are going to be from the past two weeks since I've been out the past two weeks and the first one that caught my attention was a new innovation around AI for regression testing that I think you should definitely know more about. And the first one that caught my attention was how Catalan just announced True Test Beta, which is an automated regression testing solution leveraging AI to generate and maintain regression tests by monitoring user activity. So after reading this, here are three things that it can help you right away with. The first one is how to visualize and understand how your app is truly used in production, which is a cool feature. And TruTest captures real users interactions and generates a user journey map that quality engineers could then use to prioritize and guide their testing efforts. So it's almost like a shift right type of effort getting actual data from your users and then using that data and then bake it in shift left when you're creating a test. So really cool. The second way is no more guesswork on what test because True Test automatically generates regression tests reflecting what users do in your production application. And the third way is it helps create zero maintenance regression tests because it allows quality engineering teams to focus entirely on in Sprint new feature testing and and stop spending time and effort creating and maintaining regression test. So definitely something you should try for yourself. You should join the beta program and learn more about TruTest by using the link in the first comment down below. Give it a try and let me know your thoughts. Also a quick reminder, the Kobiton Mobile Testing and Experience Summit is happening tomorrow on October 24th. It is going to be a mobile centric development and testing summit. As mobile becomes more the go to channel for enterprise operations to reach their audience, it's more important now than ever to ensure your mobile apps are being released faster while improving quality and engagement. And with the Summit. They're going to bring together some of the best and brightest minds in the mobile application development space together. Also, I'll be doing the closing keynote on automation awesomeness 260 actual affirmations to improve your QA and automation testing skills. I'm going to share with you the top themes I found after interviewing over 500 automation engineers, including mobile testers, on what they found helped them succeed with automation. You don't want to miss it. If you haven't registered yet, register now. Even if you can't make it, I recommend you register because you'll get access to the recordings after the fact as well. Look forward to your questions and hope to see you there. So we covered some news articles on large language models in the previous show, but this week I want to share with you a real world example. So Morali, in a recent blog post, unveils how large language models are changing the game in product development. So Ask ui, which as you probably know is a UI automation tool, utilizes large language models to translate natural language commands in its domain specific language, making user interactions more intuitive and accessible. So ask UI's integration with large language models goes even further by automating workflow creation, enabling users to define actions, and using screenshots. And this post also points out that the synergy between Ask UI's inference engine and large language models allows for automatic detection of UI elements and mapping of user clicks to specific actions, streamlining the user experience. Also, it points out that with open LLMs like Llama 2 has facilitated the creation of domain specific chat bots adapt to providing users with precise information and assistance, overcoming the constraints of input text limits. I think we'll see more and more of this fusion of large language models and product development tools like Ask ui, and you can read more about it in the first comment down below. So pretty much over the past year we've talked more and more about AI, and especially generative AI. But is AI a double edged sword in test automation? Let's find out. So this article is by Jason Arbon and it goes over things like how it in the world of test automation, the allure of AI, particularly models like ChatGPT, is undeniable. But Jason warns to avoid deceptive simplicity and potential pitfalls. And while a lot of AI APIs appear straightforward, automation engineers often encounter an overwhelming array of text inputs and unexpected outputs. So Jason highlights the baby talk phase where engineers request AI to generate test code, only to be met with the most generic and unhelpful response. And the lack of context and specifically in AI generated code often leads to more work for the engineer than coding manually, and he goes over the struggle extends to the over communication phase, where engineers attempt to provide exhaustive context to AI prompts. Yet the unpredictability of AI responses and the linear compound of issues for complex test cases raises questions about efficiency and reliability. Jason also touches on the flakiness of AI in this article, and Jason touches on a bunch of other ways that AI can be unpredictable and unreliable. However, Jason is a big proponent of AI and test automation as well. You just need to be aware of its pros and cons and realize it's not a silver bullet. It takes both a tester and AI in order to really make things work. It's not AI replacing a tester. So RPA type tools have entered the AI arena as well. So let's check out UiPath's latest feature so UiPath recently unveiled autopilot, which is an AI assistant set to redefine business automation. And Autopilot, powered by generative AI and a suite of specialized AI models, promises to transform paper documents into business applications with a single click. And the Chief Product Officer Graham Sheldon at uipac mentioned that Autopilot's capabilities to create robust automation workflows from users words, screenshots of PDFs make a significant stride in a one click digital transformation. And the AI Assistant's versatility extends to business analysts, developers and automation testers, offering a simplified task execution and unveiling new automation opportunities. And as Jason pointed out, there's certain things AI you need to be aware of. UiPaths added an AI trust layer that ensures data security and responsible AI usage, instilling confidence in organizations to integrate AI capabilities seamlessly and talking about AI assistance, here's another use of AI with DevOps. So Cuba just unveiled a new enhancement of a ChatGPT like AI assistance and the Cubia's copilots are engineered to interact natively with any DevOps tool, a feature that promises seamless communication and collaboration among developers across different or various engineering platforms and living with popular collaboration tools like Slack and teams. Cubia's copilots are adapt at provisioning resources, triggering continuous integration and development jobs, and generating cloud cost usage reports through simple user commands. The AI Assistant is designed to tackle repetitive tasks that boosting the productivity of DevOps teams, and this is just a great example of how more and more businesses are grappling with complex operational challenges. And using AI assistance stands as a way to help promise to reduce DevOps bottlenecks and improve developer velocity, something we're definitely going to probably see more and more of in the upcoming months and years. So what other tools can help boost developer efficiencies and cut costs? Well, Hashicorp just announced a new Terraform feature that's going to help do just this. So Hashicorp has just unveiled a suite of new features that help enhance developers velocity, reduce errors and cut infrastructure costs. The new offerings include a Terraform test framework that aids in producing high quality modules and test integrated modules published to streamline the testing and publishing process. So developers can now enjoy enhanced editor validations in Visual Studio code, making error identification and resolution a breeze. Their introduction of Stack simplifies the provisioning and management of infrastructure at scale with ephemeral workspaces optimizing infrastructure spending by scheduling automatic destruction of non production resources. And to me this is just another great example of innovation to help bolster developer productivity and efficiency for sure. All right, for links to everything of value we covered in this news episode, head on over to all the links in that first comment down below. And while you're there, make sure to check out our awesome sponsor applitools free account offer and discover to take your automation testing to the next level. Leveraging Visual AI So that's it for this episode of the Test Guild News Show. I'm Joe. My mission is help you succeed in creating end to end full stack pipeline automation awesomeness. As always, test everything and keep the good. Cheers.