371: AI Testing Needs to Mature

Yeah, that was a good point.

I remember back in the day when organizations were shipping software every 18 months or two years, testing may have not been seen necessarily as a bottleneck because this was before the era where everything had to go faster and faster and out the door.

You know, so testing had its place and everybody accepted the fact that it needed to be validated before it was sent out.

Then, of course, we got to the point now with AI generating all that code and humans clearly can't keeping up with that and the AI doing that.

So what what do you folks see or recommend to your clients as a way to, you know, help testing and QA keep pace with those things?

Yeah, well, I mean, I think certainly would start with, you have to figure out ways to create automated verification and build that into your pipeline.

I mean, that's really critical for you to be able to scale effectively.

And so whether you're You know, writing tests from a human standpoint or AI-assisted tests, you've got to ultimately feed that into a pipeline.

And you've got to be able to accelerate that as part of your delivery process.

And one of the things we recently did a survey, and one of the stats that I thought was really kind of interesting is that at least or about 69% of U.S. software experts now say that AI writes or accelerates at least 41% of their code.

And so that's a significant amount of code that we see being written.

But then testing is still happening too late.

Only about 46 are actually validating that specification before AI even generates the code.

So if you think about kind of that development lifecycle, it's certainly in a situation where we're trying to figure out how do we accelerate that testing lifecycle, get it done earlier.

Obviously, there's a number of different ways to do that, but certainly being able to automate or tie that into your CICD process is critical.

Yeah, I would think that using AI to write tests as it's writing code, that notion of shifting left, but actually building it in as a tier one player in the software development lifecycle seems to be critical.

And I think organizations that aren't doing that are going to be struggling with making sure they can get the application out the door quickly and have it mostly validated.

Yeah, that was a good point.

I remember back in the day when organizations were shipping software every 18 months or two years, testing may have not been seen necessarily as a bottleneck because this was before the era where everything had to go faster and faster and out the door.

You know, so testing had its place and everybody accepted the fact that it needed to be validated before it was sent out.

Then, of course, we got to the point now with AI generating all that code and humans clearly can't keeping up with that and the AI doing that.

So what what do you folks see or recommend to your clients as a way to, you know, help testing and QA keep pace with those things?

Yeah, well, I mean, I think certainly would start with, you have to figure out ways to create automated verification and build that into your pipeline.

I mean, that's really critical for you to be able to scale effectively.

And so whether you're You know, writing tests from a human standpoint or AI-assisted tests, you've got to ultimately feed that into a pipeline.

And you've got to be able to accelerate that as part of your delivery process.

And one of the things we recently did a survey, and one of the stats that I thought was really kind of interesting is that at least or about 69% of U.S. software experts now say that AI writes or accelerates at least 41% of their code.

And so that's a significant amount of code that we see being written.

But then testing is still happening too late.

Only about 46 are actually validating that specification before AI even generates the code.

So if you think about kind of that development lifecycle, it's certainly in a situation where we're trying to figure out how do we accelerate that testing lifecycle, get it done earlier.

Obviously, there's a number of different ways to do that, but certainly being able to automate or tie that into your CICD process is critical.

Yeah, I would think that using AI to write tests as it's writing code, that notion of shifting left, but actually building it in as a tier one player in the software development lifecycle seems to be critical.

And I think organizations that aren't doing that are going to be struggling with making sure they can get the application out the door quickly and have it mostly validated.