Machine Learning in Test Automation with Oren Rubin
What if there was a way to make your tests more reliable with less manual maintenance involved on your part? Something like a record and playback solution with built-in machine learning that creates self-improving, stable test runs? Sound like science fiction? Well, listen up and discover the second wave of AI-powered test automation with the creator of TestIm.io Oren Rubin. Oren shares his 20 years of experience in creating innovative solutions like his latest creation, Testim. In this episode, you’ll discover how Oren believes machine learning can help speed up the authoring, execution, and maintenance of your automated tests.
Oren Rubin is CEO of testim.io. He has over 20 years of experience in the software industry, building mostly test-related products for developers at IBM, Wix, Cadence, Applitools, and Testim.io. In addition to being a busy entrepreneur, Oren is a community activist and and the co-organizer of the Selenium-Israel meetup and the Israeli Google Developer Group meetup. He has taught at Technion University, and mentored at the Google Launchpad Accelerator. Connect with him on LinkedIn and Twitter.
Quotes & Insights from this Test Talk
- I think that every every challenge that you see that there's currently a challenge you're looking for. Wait a second what's the newest technology that can help me help me solve this because I'm pretty sure that someone 20 years ago also tried to solve this. And if they were smart people try to do that for example in Mercury 20 years ago. And when you started looking at what how can new technology help us. This is this is this is where you have a better usually you have a better starting point than 10 people 20 years ago. Even though I'll give you another example was that a company called Applitools in our starting point and said Oh that's impossible doing visual validation because Mercury tried to do that 20 years ago. We say oh let's use computer vision with the technology. And also it was really improving. And we said oh let's try new things. And our starting point was actually much better than Mercury.
- Recently at the latest Google conference they announced a few of their products. What I liked was that the CEO mentioning the first statement was “I think Mobile first is done and we put it aside and now it's a AI first.” And they show that everything right now is starting to be backing up by AI and machine learning
- The machine learning algorithms are well known for lots of years. The only thing that we have right now is the improvement that we have is in the way that we use them. These algorithms now a thousand time better then they were just a few years and more importantly are now applicable.
- First of all I want to say that it's it's not going to help the person. People will not start using machine learning. It's mostly going to be something that aids some tool that's going to be backed up by the machine learning. So that's something that takes a lot of data and learns from that it actually helps and improves and build a model based on that data. So I'm not sure right now that some some testers are going to say I'm going to now start using machine learning to plug it into my selenium test. I think that most of you will see that we usually see vendors and other that provide full platform they might use it at the machine learning to take the information that they're learning and actually improve.
- One of the bad things about record playback was the instability of the locators. It would record something and because it didn't have a lot of knowledge the locate there was actually much worse than any locator that the developer could think of because the developer has a lot of knowledge on the application. But if he'd taken up an application now and the computer will start learning about your application start understanding relationship between the parts of the DOM and actually more than that what if you started learning about changes throughout time. Think about how our locators right now are very are very static. And they assume that application will not change and its only one way of finding an element. But if you actually ask the machine, what a second, why don't you look at the last few months. Once you started learning and see how the application changes then you can ask it OK why don't you give me a better locator that can actually work better? What if you tell the machine, why did you not just give me one locator, can you give me a hundred locators so you can actually check all of them. Because right now it's super fast. It takes milliseconds to find an element that. So I can try a hundred locaters. You can then ask the computer, Ok now that I have a hundred different locators I can try them all. Why. What if I can say wait a second. If some of them give different results. Well, obviously they all get the same result same element you say oh that's easy. That's that one. Well, what if some of them couldn't find anything. Or something different results. And you're starting to have to start thinking Who's the better locator? Who should I listen to? And with those kinds of things, a machine can do much better than humans. They can take up an enormous amount of data information and extract out of that information and have results which are much better much accurate than a human.
- So you can have automatic maintenance. I think the nice part is that anyone can start using that. That means that the support team can also start using that when they open bugs they open tickets with automatic test automation. So this is the part that I think that can also be very helpful. And then you have the entire team helping out not just a secluded QA everyone has to now be part of the QA. The quality of the product is one of the most important things.
Connect with Oren Rubin
May I Ask You For a Favor?
Thanks again for listening to the show. If it has helped you in any way, shape or form, please share it using the social media buttons you see on the page.
Additionally, reviews for the podcast on iTunes are extremely helpful and greatly appreciated! They do matter in the rankings of the show and I read each and every one of them.
Test Talks is sponsored by the fantastic folks at Sauce Labs. Try it for free today!