How do you Test an AI ChatBOT | Learn Playwright Before It's Too Late!

[00:00:02]Hey everyone. So in this video we are going to see the interview type of question that can be asked to you from the AI perspective, right? So how would you test an AI chatbot? Okay. So these kind of questions have been coming in the interview these days, right? Testing an AI chatbot. How would you do? What would be your approach, right? So we will see the answer for this particular interview question, but before that let me tell you about a free webinar that we are going to do on the coming weekend, that is 25th of July, 8:00 a.m.

[00:00:35]from 11:00 a.m. IST onwards, right? So uh as you can see, more than 200 to 250 plus people have already enrolled and we are left with last 120 seats. So you can enroll for this particular webinar today. I'll be sharing the link in the description of the video, so that it becomes easy for you to Okay. Now, how do you test an AI chatbot?

[00:01:05]Okay. So there are different things with respect to AI chatbot, but how do you test it? So first of all, when you perform an application or testing on a normal application, Okay. That is different from how you perform the testing with AI application.

[00:01:30]Okay. So testing on a normal application is different from testing the AI application because in the AI you are knowing the question or the behavior and the judgment, right? But it's not just a fail pass and fail logic. See, normal testing what happens?

[00:01:51]It's a pass and fail logic.

[00:01:55]Okay. But here with respect to AI we don't have to test just the pass and fail logic. We have to more focus on the behavior and the judgement.

[00:02:08]The precision.

[00:02:12]And the accuracy by which the response is given.

[00:02:17]Okay, so any chatbot application will respond, right, in its own judgement with its own decision behavior. So you have to test that.

[00:02:27]Okay, so from functional testing point of view, what you should tell in an interview is, right?

[00:02:34]You have to check the intent recognition.

[00:02:40]Okay, what is intent recognition? Intent recognition is nothing but you have to check if the bot correctly identifies what the user wants. Okay, for example, I want to cancel my order versus I don't want this order anymore versus cancel, please should all map to the same cancel order intent.

[00:03:01]Right? So when I say I want to cancel my order or I'll say I don't want this order anymore or cancel, please. So the intent of these three statements is the same. So ideally these three are your test cases or you have to prepare test data in such a manner that you can verify the decision-making or the judgement of your chatbot. Right? So this is one thing. So next thing that we will see here is your entity extraction.

[00:04:10]>> So, what is entity extraction? What does that mean? Okay, so you can take 4 or 5 seconds to think on this.

[00:04:19]This is nothing but it is to check if it correctly pulls the structured data from the messy input.

[00:04:34]Okay.

[00:04:36]For example, when you tell to a chatbot that book me a flight to NYC next Friday for two people.

[00:04:55]Okay. What does this mean? It should extract destination is equal to NYC.

[00:05:02]Okay. The extraction details are right here.

[00:05:05]One is your uh destination.

[00:05:12]Another is your passengers.

[00:05:21]Okay. Another is the date date of travel, right? That is next Friday.

[00:05:27]So, this is nothing but this is known as entity extraction in which what will happen?

[00:05:33]You will be working you will be checking your chatbot is responding and it is able to extract the things or not, right? And then you have to check for the fallback handling as well.

[00:05:49]What is fallback handling? It's nothing but if you give any any something like this, okay, any data like this. So, what is the meaning of this to a bot? So, it should gracefully say, "I did not get that." instead of hallucinating an answer, right? So, very important thing is you have to test your chatbot in this particular manner. These are few of the scenarios, few of the invalid scenarios. Like this, there are multiple other scenarios that you will have to think on. Okay, and you will have to include them in your testing. So, ensure that you test a chatbot in this particular manner, right? Because these kind of questions you would get in an interview, and if you're not able to answer, then you will not be able to proceed to the next round. Now, you might be feeling that in my company these things is not done, but in other companies people have started using AI and develop an AI, design an AI, test an AI. So, you should be able to give these answers properly. You should be able to make sure that you are giving the answers because if you won't be able to answer, then they'll come to know that you are not testing an AI itself, right?

[00:07:01]So, test an AI, have that hands-on experience, and give the answers in this manner if you want to get selected in an interview.

[00:07:08]Right? So, that's it for today's video, and last but not the least, I would like to tell you that we have free webinar which will be happening on the upcoming weekend, 25th of July. So, seats are getting filled. So, please do enroll today.

[00:07:26]And in this webinar, you will also see how agents can do the work of automation testing. How they can do the work of creating the test cases from the scratch within testing this language. And apart from that, it can also run the test case, update the test case, maintain the test case, and I'll also tell you about the job market.

[00:07:49]What are the skill set they are looking for? What you should do in order to stay updated.

[00:07:55]Right?

[00:07:58]So, these are the things that you should be looking forward to, right? So, that's it for today's video. Thank you so much for watching this video and stay tuned for more.