Transforming Application Testing with AI – Esko Hannula, Copado

This is texturing TV. Hey guys. Thanks for the throw.

We're here with Esco hanula who's senior vice president for product management at capado and we're talking about Ai and testing there's been a lot going on here. They're advances are rapid and we're gonna Jump Right In Esco. How you doing?

Welcome the show. Hi, Mike, doing good. How are you?

I'm well, there are a lot of phases and testing. So where exactly does it make sense to use AI these days what should people be expecting and just what is the level of automation that I can get to? I think you can use AI for almost anything you would do in testing, but obviously the greatest opportunities right now are in in test case design and this generation.

which is kind of exciting because when people talk about this automation, they usually mean automated execution of tests that the human being decide. So what's really different now is that AI can decide the tests with assistance of human being or human being with the assistance of AI whichever way you like? Intense the case then it is one thing to be able to automate tests that already exists using AI but the problem is always been the creation of the test is difficult sometimes or nobody has the time to go do that.

So if we can rely on AI to design the test more and will we actually test more as a result because the issue is always been kind of twofold one is we didn't have enough people to design the test and then we ran out of time to execute the test. I think so. I I certainly think so.

Yeah this thought of my son in its original form. It saved a lot of human labor, but in a way that's execution is something that the robots should have been doing all Since the very beginning it's it's not something very you need to human being you need a human being or intellectual part, but Humans Beings are well, it just scarce resource also for this design. So if we now have a new machine that is able to design meaningful tests.

It will mean we have more tests to run it will eventually mean we will find more bucks, which means we need means we need the human beings then the fixed the software. Well, maybe we'll get to the point where the AI can also remediate the software because a lot of the bugs I understand at least are fairly common. I bet somebody's working on that that already and in fact, it's not that far.

It's that Ai and actually right to code. We have a lot of different types of AI these days the most popular at least in terms of the imagination. Is this notion of chat GPT?

What's your sense of generative Ai and its role specifically in testing versus what we have been using to automate testing. Yeah, it's a cheap PT really changed everything and it's kind of amazing those Technologies underlying tech TV. They had been around for 25 30 years at least.

but In addition of those algorithms we need it. The huge amount of data on the internet and a huge Computing capacity that can actually build those models using using the data. Data on the internet and check if the typical kind of popularized.

AI so that every everyone all of us can understand what AI is capable of and that almost like automatically race the level of ambition in AI applications now prior to generative AI we used AI to compare things. So show a picture is this a cut? Yes, 95% probability.

It's a cat now. We generative AI we can ask make me a cat. And and you you get it and really if it can generate.

English or French or whatever Language and an answer questions. It's obvious. It can also generate selenium or or some other tests test Scripts.

What is your sense and what the impact of this isn't going to be we have been trying to push testing further left towards developers. Do you think this will make it easier to achieve that goal or will it still need dedicated testing teams from time to time? What can I push left and what stays right?

That's a very good question. I believe we still need dedicated testing teams. Probably those medicated testing teams will be spending their time or something else.

They do today because AI will be will be helping them. developers mostly test at the level of code and developers can certainly make use of AI and and do a lot to use AI for many of those tasks they used to Play with with testers. but so far, it's an every attempt.

To make developers test successfully no matter what tools are available. Has been inadequate and it's probably because the same breaks that create the code. Should never create the best.

And of course with AI. One could argue that there are different ways. No working on the best.

Yeah, that's roughly equivalent to asking the fox the guard The Henhouse, right? Essentially if I test my own code. Very much so very much.

So and it's not necessary the developer. Wouldn't know how to test or wouldn't like testing some of the don't know some of they don't like but it's just that if you build something. It's really against the nature that you would then try to destroy it.

Which is essentially the top of the tester. Do you think as a result of all this that the current state of application security might also improve as well because one of the things that we're testing for should be testing for are all these flaws that show up in applications and this has been a problem for years now, but might this suddenly all get a lot better. it might that's a very good very good point.

both in say security testing because if the AI learns how to how to test functionality it will also learn how to test security but I bet we can also create better code scanners with AI that detect typical security problems in the total level. What is your sense then ultimately of well, we're talking a lot more about liability of software these days. I don't know if you saw the US National Security strategy document and there was a lot of talking about liability in that case and it's always been difficult.

It's a whole people accountable for application security. Do you think the tools are advancing to the point where well, maybe it does become feasible to kind of hold people more accountable for the quality of this level of security included in their applications. We should certainly hold people and companies more accountable for for security of their applications, but there's a limit to what is reasonable.

I mean security measures we take them against funnel ignorant or or careless users and we take those measures against malicious users and an even criminals. and well in all other areas of life those who do wrong things are always a step ahead of those who write to protect people against us from things and there's no reason to believe that same is happening also in software. Do you think also that one of the side benefits to this is that?

One of the issues that you hear a lot about is that organizations have a lot of turnover in their application testing teams. It's sometimes a thankless job and no one thinks you're for doing it. Right and they're yelling you when you do it wrong.

So people kind of decide that, you know, maybe I want to go do something else in my life. Do you think that if we augment those folks with AI we might see less turnover because the job itself will just become that less tedious. oh, definitely, and I'm I'm happy to say that the appreciation of software testing as a profession has improved a lot or past couple of decades, but it's it's still In too many places considered a kind of second class software job, which is student because it's intellectually very challenging job.

Actually when you do it, right and when you become good at it. but what we will see with the increased use of AI insult for testing is that anything that the machine can do will be done by the machine and the human beings that will be needed for testing or probably on the sharpest side of the of the of the testing testing teams. So I believe AI will just improved appreciation of the professional software testers.

Do you think the relationship between developers and testers will change because a lot of times it's adversarial and maybe developers will start viewing this as a little more of an aid or a help because it won't feel like they're being constantly called to the principal's office to get in trouble. It's going to be more like Hey, we're seeing this in context while you're writing the code. So let us tell you about this now versus three weeks from now when you move down to some other projects.

I certainly hope that is the case and we see that happening in many great software organizations already today. Yeah, so let's go. What's your best advice to folks?

I mean, how should they proceed? I think a lot of people are intimidating a little bit by AI. So where do you get started?

Well, the first piece of art was if you feel intimidated forget it AI is coming anyway in all areas of life. So there's no point going aggressively against it or ignoring. Let's learn what it can do for it do for us.

Let's explore with it. And let's see us as an opportunity. It will certainly change the way we work and the way we build software and that change will happen no matter what each of us decides to do.

So, let's take an open mind open mind see AI as an opportunity and start exploring with it. All right, folks here heard it here. The AI parade is coming.

You can either let it run you over or you can turn around and get in front of it and Lead it. I think the you want to go for Plan B there Esco. Thanks for being on the show.

Thank you so much. This was great fun. All right back to you guys in the studio.