🎙️ [Episode 37] Exploring the Power of ChatGPT and Intelligent Automation

Shaun Leisegang:

Hey there, thanks again for joining us on this episode entitled The Future of Work, Exploring the Power of ChatGPT T and Intelligent Automation. Well, it seems like you can’t go anywhere at the moment without hearing about ChatGPT and it certainly is taking over the internet at the moment. So I thought it would be really good to do an episode on all things ChatGPT and Intelligent Automation. So let me take you through some of the amazing things that I have found. First up, what I’m going to do is I’m going to go and head over to the OpenAI website and what I’m going to do is I’m going to go ahead and log into ChatGPT. Now what this does, it opens up a chatbot type window where I can obviously ask a ChatGPT a number of different questions and it’ll obviously feed back the answers to me in a chatbot type format.

So what I’m going to do is I’m going to start off pretty simple and the first thing that I’m going to do is I’m going to actually enter into the chat and I’m going to say, Explain ChatGPT to me in simple terms,” and I’m going to hit enter. Now what happens is as soon as you enter the question, what happens is ChatGPT goes away and then it starts to go through and obviously write out some text for me. So what we can see happening now, is ChatGPT is writing back to me and what it’s saying is, “ChatGPT is a type of language model that is trained to generate human-like text. It uses deep learning techniques and specifically a type of neural network called the transformer to learn patterns in large amounts of text data and generate new texts that is similar to the examples it was trained on. It can be used for a variety of natural language processing tasks such as language translation, tech summarization, and conversation generation. In simple terms, it’s a computer program that can understand, generate and respond to natural language.”

And it’s pretty amazing that I’ve just said a simple term like explain ChatGPT to me in simple terms and it’s automatically understood the context of the question that I’m asking and gone through and generated me a specific answer. So next up what I’m going to do, is I’m actually going to just extend on that question. What I’m going to do is I’m going to type into the window, “How does ChatGPT relate to Intelligence Automation?” And again, what ChatGPT is going to go away and do, it’s going to go figure out the context of my specific question and then it’s going to start generating me a specific answer.

So here it comes and it’s come back and it’s said, “ChatGPT and other language models like it can be used to build intelligent automation systems, which are systems that complete tasks that typically require human intelligence to perform. For example, ChatGPT could be used to build a chatbot that can understand and respond to customer inquiries in a natural and human-like way, or to generate automated responses to emails. It could also be used to system tasks such as text summarization, language translation, and even text generation. By automating these kinds of tasks, intelligent automation systems can improve efficiency, reduce the need for human labour, and improve the overall experience. For example, using ChatGPT to assist in customer service can help respond to customer needs and inquiries faster and more efficiently. It can be used in various domains such as healthcare, finance, retail, customer service, and many more to automate repetitive task process automation and decision making.”

That is a pretty amazing response. Just by that simple question, it’s gone away and figured out all of that content and how ChatGPT relates specifically to intelligent automation. So that’s a really good start, but what I’m going to go do now is I’m going to go through and I’m going to actually give it something a little bit more intelligent to do. So I’m going to say to it, “Write me a blog post on why companies should investigate ChatGPT and Intelligent Automation in 2023. Again, what that model’s going to go do, it’s going to go away and it’s going to go and create me a blog post around why companies should actually investigate ChatGPT and Intelligent Automation in 2023. So here it goes and it’s already starting off and it says, “As we move further into the 21st century, businesses are facing new challenges and opportunities in the realm of technology and one of the most exciting areas of development is the field of Intelligent Automation, which combines machine learning and artificial intelligence to automate tasks that were previously performed by humans.”

Now it’s going to go on and on and it’s going to actually generate me an amazingly good article or blog post on why companies should investigate ChatGPT and Intelligent Automation in 2023. So let’s let it go through and generate the full article for us. All right, so just while it’s going through and generating the article, I’ll just read it out for everyone. “So one of the key technologies driving the developments of intelligent automation is a natural language processing or NLP, which enables computers to understand and generate human language. And one of the most advanced NLP models available today is ChatGPT. There are several reasons why companies should investigate ChatGPT and Intelligent automation in 2023, and these are improved efficiency. So the ability to automate repetitive and time consuming tasks that can free up human employees to focus on more important and value adding activities. For example, using ChatGPT to handle customer service inquiries, can help businesses respond to customer needs more quickly and effectively.”

“Number two is all about reduced costs. Automation can help reduce labour costs as well as improve of accuracy and reduce errors. This can be particularly beneficial for businesses operating and in competitive industries. Number three is all about improved customer experience. With ChatGPT, companies can create more natural and human-like interactions with customers. This can help improve customer satisfaction and loyalty. Number four is all about better decision-making. So automation can help companies access process and analyze large amounts of data in real time. This can improve decision-making abilities, enabling companies to make better use of the data to make more informed business decisions. Number five is all about market differentiation. So by having the ability to use intelligent automation system, using ChatGPT gives you a competitive edge over the players in the market. It can help you in providing better customer service, effective decision making and efficient automation of repetitive tasks. In conclusion, ChatGPT and intelligent automation are powerful technologies that have the potential to transform the wave businesses operate. As we move into 2023, companies that investigate and embrace these technologies will be well positioned to stay competitive and. Meet the demands of the 21st century marketplace.”

Truly amazing that it’s managed to generate me an amazing blog post with some very, very informative content on the topic of ChatGPT and Intelligent automation in 2023. Now the next thing that I did is I wanted to say to ChatGPT, “Well, okay great, that’s a great blog post, but what I want you to do is include in the above that rapidMATION are an Australian-based company that specializes in intelligent automation and if anyone would like to discuss their intelligent automation needs to reach out to us at hello@rapidmation.com. So let’s see what it does. It’ll go away and regenerate that specific blog post for us.

All right, and there it’s all done. So it’s created the same blog post for me, but it’s actually appended to that and it’s said, “If you’re in Australia and looking to investigate and embrace these technologies, rapidMATION, an Australian-based company that specializes in intelligent automation is ready to help you with all your Intelligent Automation needs and you can reach out to them on hello@rapidmation.com. Truly amazing that it’s built on the context of that full conversation and included me some different piece of information. Next up what I wanted to do there is I wanted to say to it, “Okay, well what I want you to do is go and create me a title for this blog post.” So inside the chat window I say, “Create me a title for this blog post.” And what it’s actually said is, it’s created a title or recommended a title that says, “Exploring the benefits of ChatGPT and Intelligent Automation for Businesses in 2023.”

Pretty amazing. What we’ll do is we’ll publish this blog post out on the rapidMATION blog, so if you head over to rapidmation.com/blog, you can go check out this blog post that’s been generated by ChatGPT. So next up what I normally do is I normally actually go through and I promote these blog posts via social media. So what I’m going to enter into the ChatGPT window is, “Create me a social media post that promotes the blog post,” and what it’s done, it’s come back with an answer that says as follows. It says, “Are you ready to stay ahead of the competition in 2023? Check out our latest blog posts on the benefits of ChatGPT and intelligent automation for businesses. Don’t miss out on the opportunity to improve efficiency, reduce costs and improve customer experience.” It’s even put in some hashtags for me. #ChatGPT, #Intelligent Automation, # business efficiency and # business growth.

Truly, truly amazing that I’ve been able to create this content using ChatGPT. All right, so that’s some of the examples that I’ve seen. I then came across a LinkedIn post from a gentleman by the name of Stuart Chambers from UiPath where one of his UiPath colleagues, a gentleman by the name of Mick Taylor, showed a demo of how UiPath Test Suite and UiPath could work together. Now, I must be honest, I was blown away by what it managed to do and I didn’t actually believe that it was real to be honest. So what I had to do is I had to go and actually test it out myself. So really what the context of the demo was, is that Mick had created… Inside the UI path world, what there is there is a number of different applications that they have that they use for different demonstration purposes.

And in this example what they did is they’ve got a utility called uibank.uipath.com and it’s really just a webpage where you can go and actually input different pieces of information and you can build automations around it. Now on the screen you’ll see there’s a loan application and really what it’s got, it’s got a few different fields. It’s got an email address of the requester, the loan amount requested, a loan term, which is a dropdown list box, the current yearly income and the age. And then it’s got a button that says submit loan application. So in this demo, what the team was showing is how we could actually generate some test data and some test cases for this specific screen using ChatGPT. So I’m going to head back to ChatGPT here and what I’m going to do is I’m going to say in the chat window, I’m going to say, “Please go ahead and generate me five rows of CSV test data for the following website.”

And I’m going to put in the website URL, which is uibank.uipath.com/loans/apply. And I’m going to hit to enter and we’re going to go see what ChatGPT goes and does for us. Now in order for ChatGPT to understand this, what it’s got to do, it’s got to go and obviously look at that specific webpage, figure out all the different fields that it has on that form and then go generate me different pieces of test data. So what you’ll see coming back in the ChatGPT, window is we’ve got it generating me five lines of data. So it’s got name, email, phone number, loan amounts and credit score and what it’s got, it’s got John Smith, a John Smith email address, a phone number, a dollar amount and a credit score. And it’s generated five different records for me. Now that is truly mind blowing that it can go away, investigate a website and then generate test data for me that I can go through and answer on that website.

Now what we are going to do, is we’re going to say here that it’s obviously entered some loan amounts and some credit score information, but what I’m going to do is, I want to add some more information into this test set. So what I’m going to do is I’m going to type into the ChatGPT window, add the following fields to the previous test data, and I’m going to add in age, yearly income and loan term. So basically what it needs to do now is it needs to go and actually append different pieces of information to this test data. So what ChatGPT is currently coming back, it’s completely understood that request, it’s listing out those five previous test records that it’s done for me, but it’s included an age, it’s included a yearly income and it’s actually included a loan term for me.

Again, absolutely mind blowing that it can figure out all the context for me and that it’s generating me all of these different pieces of test data. Now I thought that was actually really, really amazing. But the next thing that Mick did at UiPath, is he then entered in to the ChatGPT window. He said, “Okay, go and generate me five test scenarios for the website, uibank.uipath.com/loans/apply,” and what ChatGPT has gone away and done now, it’s taken all of that context of the form and the data that we need to input there. It’s given me some test data and what it’s feeding back in the window is it’s giving me five different test scenarios for this specific webpage. Number one is, a user fills out the loan application form with valid information including all required fields and successfully submits their application, which is a hundred percent a valid test.

Second was, a user attempts to submit the loan application form within incomplete information and receives an error message prompting them to complete all the required fields. Again, a completely valid test and it goes on and on and on. Now the next thing that we can actually do is we can actually expand these first five tests. And what I’ve put into the test window now is, “Go ahead and generate me five more test scenarios. So it needs to understand what it’s recommended to me before and it’s going to go through and it’s going to go actually give me five more test scenarios for this specific example. Now again, completely amazing for anyone who’s been involved in the test space. Generating test data or even test cases is often a tedious process, something that’s really, really important, but using the power of ChatGPT and something like UiPath Test Suites, we can go through and we can generate all of this test data and all of these different test scenarios, which is really, really, really amazing.

All right, so that was a really, really great start. What I then did is I wanted, wanted to go investigate some other bits and pieces of what other people out in the industry have done. And what I did is I then headed across to one of our other partners called Nintex. And for those of people who may not be familiar with Nintex, Nintex is a process automation platform that has a number of different pieces from process mapping to automation to reporting and beyond. And what I came across is I came across a really, really cool integration component that a gentleman by the name of Deon Smit created. And Deon created a Nintex integration into ChatGPT and to DALL-E. Now for those of you who may not be familiar with DALL-E, DALL-E is also something from OpenAI. And what DALL-E actually does is, DALL-E is an AI tool that can generate unique images and content based on user prompts.

So it can be or used to automate tasks such as creating marketing materials, product design and visual content creation. It’s truly, truly mind blowing. So what Deon, this example, I encourage you to head over to gallery.nintex.com and there’s the post, Nintex Automation Cloud Meets Artificial Intelligence. There’s a really, really great video that Deon has done, showing this in action, but I’ll just take you through a whirlwind tour of what he actually did here. So the first thing that he did is the whole scenario was around a leave request and he built out some integration components using the OpenAI API to ChatGPT and to DALL-E. And what he did is, he allowed a user to submit a leave request and inside this leave request, what he did is he captured some details. So who the requester was the start and end date, the type of annual leave. And then he allowed the user to input some comments on their annual leave.

Now the first thing that he did is he specifically used some pretty bad English language, incorrect tenses and spelling mistakes. And actually the comments read or read, “I had some issues,” which issues were spelt incorrectly, “with the cattle on my farm, which,” which is spelled W-I-T-C-H rather than W-H-I-C-H, “need,” N-E-D, rather than N-E-E-D, “my immediate attention.” And both immediate and attention were spelled incorrectly. So he’s purposely gone through and he’s actually used some really bad English with some incorrect tenses and some spelling mistakes.

And then what he’s done is he’s exposed a little dropdown list box that says, “What is your manager’s preferred language?” And in this case what he’s done is he selected a language called Afrikaans and why he used the language called Afrikaans… Afrikaans is a language that Deon speaks and what he wanted to do was he wanted to then go through and firstly allow GPT-3 to correct the comments and the language and the tenses. And then what he was going to do is he was going to use whichever language was selected for that direct reporting manager and translate that email and all the details into Afrikaans, which again is pretty, truly amazing.

So when this workflow runs, you can go through and you can see Nintex gives you the great ability to see who did what, when, how, and why. And here you can see the full run sheet of what happened inside that workflow. So what happened inside that workflow is we reached out to OpenAI through the API and we corrected the comments to standard English. He then went through and classified the sentiments of the text, he then performed the translation from English to Afrikaans and finally he reached out to DALL-E to get an image representation of the leave request reason.

So pretty amazing to be able to do this from an Intelligent Automation process. And what it looks like in the end is what it does, it sends out a specific email and on the right-hand side we’ll be able to see the email. And for those of you who can see that email, you’ll be able to see that it says, “Hello Deon,” It said, “Deon Smit het die volgende tyd afgesoek,” and that’s Afrikaans. So it’s actually translated all of that email into Afrikaans. And what we’ll notice there is a few different things. Firstly, the email’s being translated into the correct language and it was pretty good from what I can see. He’s also gone through and formatted the dates from Zulu time to human readable date format. And then we did a sentiment classification so we can see that there was no hate or anything like that. So it gives you some true and false on the sentiment classification.

And finally, inside the email, he’s actually used DALL-E to go through and create an image and the image actually shows a number of different cattle on a farm. And that’s actually the context of what the reason was for this person going through and taking some time off. Truly, truly amazing that we can combine intelligent automation with ChatGPT to deliver some outcomes like this. An excellent piece of work done by Deon and you could go check that out on the Nintex gallery. And just as a reminder, if you wanted to go check that out, head over to gallery.nintex.com/t/nas_meets_ai and I’ll put that in the show notes as well for anyone who may want to go check that out. Again, some amazing capabilities there.

All right, but we’re not done just yet. Let’s head back across to our ChatGPT window. And I’m going to go through and I’m going to start a new chat. Now, the next thing that I wanted to check out to see what ChatGPT could do for me, is if it could generate any specific code. So what I’m going to do, is I’m going to go pop into the ChatGPT window. I’m going to ask it to write me .NET code in C# code to call the OpenAI API. And I’m going to hit to enter. Straightaway ChatGPT comes back to me and it says, “Here is an example of how you can call the OpenAI API using C# and the .NET framework.” And it’s literally given me a code window where it’s generating the code that I could copy and paste, customize with an API key and run straightaway to go through and actually interact with the OpenAI model.

Absolutely mind blowing that a chatbot figures out all of this context for me and allows me to go through and generate the code that I can use to actually interact with it. Truly, truly, truly amazing. So I guess what I wanted to do in this episode is just take you through how intelligent automation and ChatGPT could be used in combination. And the summary is that the possibilities with ChatGPT and Intelligent Automation are truly, truly endless. So I for one personally, can’t wait to see what 2023 holds for ChatGPT and how we can use this in combination with intelligent automation to take many bright ideas and turn them into automation champagne. I hope it’s been useful and a great spend the time just to see how we can combine these worlds of ChatGPT and Intelligent Automation. And as always, if you’ve got any questions, reach out to us on hello@rapidmation.com and we can explore some more ideas.

Speak soon.