SmartBear Talks | Lino Tadros - AI in Software Tes...
SmartBear Talks | Lino Tadros - AI in Software Testing
In this interview, we are discussing one of the most exciting topics in the computer-based world - artificial intelligence (AI). Especially, AI in software testing. We have invited Lino Tadros, AI & Machine Learning Architect at Solliance Inc., to share his thoughts about the current AI implementations and the future of AI in software testing. Should we expect to see robots who will execute tests? Let's watch the interview to learn Lino's predictions.
President & CEO – Events XD Corporation
AI & Machine Learning Architect – Solliance Inc.
Microsoft MVP for 15 years
TestComplete Expert for 19 years – Official Trainer and author of the Certification Exam
Author of 8 TestComplete books
Architect of major projects at Disney, NASA, HP, Microsoft and many other corporations around the world
Technical speaker at conferences for 25 years – 52 countries so far
Conducts all the TestComplete official training every month for Keyword and scripting for SmartBear
Any comments/questions/thoughts on artificial intelligence in software testing? We will be happy to discuss them - just leave a comment below.
Note: you can learn TestComplete from scratch or improve your knowledge - register for a special course Lino is setting up on June 8-9. You will get a lot of interesting and useful information!
Key interview moments:
Let's talk about AI in software testing. Many people including me imagine robots when they hear ‘AI’. Those robots are as clever as humans, and they follow the Three Laws of Robotics not to harm people, I mean that's all that we know from the movies. Should we expect robots in software testing?
It's actually a lot more than that. It's not only robotics. It's about really creating a model that can understand what people are doing, and the hectic amount of information that you can have. AI requires a lot of data, so when you give it so many patterns and so many algorithms, it can actually start predicting based on the past. So there is no fear of AI. Artificial intelligence actually is something here to stay and for the long term, and that's the future of our software world going forward, but it's not only the robotics.
How will it look in software testing? Your prediction.
I think the future of software testing with AI will be amazing. And, we've seen this with TestComplete, they just have added some new OCR based on the Google vision. The more people in the world use OCR, the system will continue to learn, and your recognition will get better and better.
The future I would love to see for testing tools is for them to be able to understand how people work. For instance, sometimes, testing websites is a little bit difficult because there are a lot of dynamic issues inside - some of the controls that are being created on the fly change names all the time. And, we see a lot of people always complaining about ‘object not found’. And, I would love to see the future of AI that will actually be a lot more intelligent about the structure of a webpage to understand where everything is, and if names change it would be smart enough to know that all the other characteristics of the controls are the same - the inner HTML, the outer HTML do this and that; the name just changed so it would change it on the fly to keep everything working correctly. That would be a very smart recognition system and a name mapping system. I think that will be the best thing for the future.
Is there a way AI can be used in order to help testers work? I mean, maybe, to decrease the number of tasks, or, maybe, to decrease the time of the test execution or something like this?
Again, by training models. I mean, the one thing is if you have a lot of projects in TestComplete, you can throw them all into a model, and the system can actually start learning how your company does testing by learning the structure of your organization project, for instance, the project suite, and understand how you like to do things, how you like to name your applications. So, AI could actually save a lot of time.
But as far as I understand this teaching process, we will need to teach the model, or, maybe, the system about our testing process. And, what I am afraid of is that this teaching process can take some time, and most of the companies won't be ready to spend the time on this because during this time they will do tests instead of teaching.
It is an investment. I mean, for instance, there are a lot of projects for AI that I work on right now whether it's for the aviation industry or for the New York bus system, for instance, and so on. And, it's not like you can give the system like a hundred flights delayed, and then it can predict for you if the next flight is gonna be delayed or not. That is not possible. You have to give it pretty much all the flights since 1970 so it would have to be in the millions of flights that happened, and then you can get a good prediction. But, a lot of people think that by giving something a small amount of data they can get a good result as far as the prediction and all that stuff. But, it’s about investment. You’ll have to put in a lot of data for the system to be more accurate.
How much do you think we can rely on AI in software testing?
I think the future of software would require AI in testing because the software being written today for the next five to ten year is actually counting on significant amount of data. And, I don't think, actually, testers will have the time to test this even with automated testing without actually using AI to generate a huge amount of data to test these applications. So, I would say to all my fellow testers in the world - it is time to actually start looking into AI and machine learning, understanding the concept of what it means, the difference between artificial, intelligence machine learning, and deep learning. And, try to understand that in the future of testing for you and for more job security, you will really need to understand how to use models to be able to train your testing platform on being able to recognize things in your applications without you having to do everything manually. I recommend everybody to spend some time on it because it's the future.
Thanks Lino, thanks a lot for this conversation. That was very interesting. Community, if you have anything to add, or if you have any questions, please post your comments under this video, and we will be happy to continue the conversation around this interesting topic. Thanks Lino, thanks for your time.
Thank you so much Tatyana for having me, it was a pleasure, thank you.