In a matter of weeks, generative artificial intelligence (AI) has moved from the bleeding edge of technology to the mainstream.
Right now, you, me, and just about anyone can use AI to create amazing artwork, narrate a book, write a blog, even create a new programming language from scratch. Imagine going back just a couple years ago and telling your colleagues you’d be able to do all of that today from your smartphone!
The launch of OpenAI’s ChatGPT—a large language model (LLM) able to generate human-like text—has been a catalyst. It has revealed to millions of people around the world the recent advances in AI and focused minds on what’s possible with these models. It’s a big step forward in the democratization of AI, providing powerful foundational models that can be used to create even more valuable tools.
With that in mind, I decided to check in with one of our resident expert robotic process automation (RPA) developers, Cristian Negulescu—and a few other friends—on some of the ways they find ChatGPT to be useful.
(Note: ChatGPT is the most accessible generative AI tool available today, but similar results could be obtained from other emerging tools as well)
A great use of ChatGPT is helping professional developers answer their questions about other developers’ code. It’s as simple as pasting the code into ChatGPT and asking for an explanation of what it does. This can help automation developers to better understand the likes of VB.Net, SOQL, JQL, LINQ, and any other coding language.
Benefit: Enable developers to expand their knowledge of lesser-used programming languages
ChatGPT can also help ease the pain of creating lots of user documentation for code. After you paste all the information related to your system into ChatGPT, you can ask it common questions users might have and use the output for your own documentation.
Soon, a more advanced version of ChatGPT or a similar generative model could even help keep user documentation updated. Imagine communicating the latest changes to your system and having the model update all the collateral needed for the new version.
Benefit: Create user documentation more efficiently, while potentially automating updates, saving developers time and effort
One of the most exciting possibilities for ChatGPT is helping even non-technical users become developers. Generative AI models allow business users and citizen developers to generate automation workflows from natural language descriptions. By this, I mean they can explain what they want to achieve and ask the model to write the code to achieve it.
This is the superpower of generative AI. A ‘buddy’ to help any level of user towards accomplishing their automation goals and realizing new potential in their day-to-day work.
Benefit: Helps non-tech users generate automation workflows from their plain language descriptions, driving citizen development opportunities
When a developer needs a set of test data—such as, marketing contact lists—with a set of specific fields (e.g., Name, Phone Number, and Address) for a specific state or region, ChatGPT can quickly generate a randomized set of fake data.
Here is an example prompt: “I created an application that will work with phone numbers and addresses. The following fields will be used in a record: Name, Phone Number, and Address. Generate 30 examples of records, specific to Washington State, USA.”
Change the prompt to your locality, or for only names starting with the letter A, and see how easily the output can be customized.
Benefit: Accelerate the creation of test data for developers by generating randomized sets of fake data that respect logical integrity rules
Let’s assume you have a sequence of code and want to create a test script for it. ChatGPT can assist and generate testing code, in languages such as Python, JSON, C or XAML (or even convert between the different languages).
Additionally, ChatGPT will also create the test data needed to cover all the test cases.
Benefit: Accelerating the creation of test scripts for workflows
Moving beyond the realm of developer tools, let's explore some of the business applications of ChatGPT when paired with a UiPath automation.
One spectacular use for ChatGPT is analyzing text and evaluating the sentiment of customer feedback on a specific product.
To demonstrate, we can provide a set of product feedback received from the customers to ChatGPT and enter the following prompt: “Please identify the sentiment of this feedback, assigning it as positive, negative, or mixed.” ChatGPT will respond, identifying the tone of messages consistently and accurately.
We could even add automation to the mix. A robot can send ChatGPT the feedback list, share the prompt, receive the answer for each message and count the positive, negative, and mixed answers. Based on the sentiment, particularly useful feedback can be processed downstream. For example, negative feedback could be automatically directed to the product development team for consideration.
For more in-depth, real-time sentiment analysis and intent extraction from customer feedback, you’ll need a tool like UiPath Communications Mining.
Benefit: Accelerate the analysis of customer feedback of a product or service
Providing good customer service, in good time, can be a real challenge for businesses. Luckily, ChatGPT and UiPath robots help here as well.
Let’s assume a customer shares negative feedback on a product. We could use ChatGPT to write an appropriate response. The whole process could even be automated from end to end.
When negative feedback is received, a robot shares a prompt, and the text of the email, with ChatGPT. ChatGPT responds with an appropriate response message, which the robot then validates with customer support before sharing the response with the customer.
Benefit: Improved customer experience and time to resolution
ChatGPT can also be used to prescreen resumes and estimate how well candidates fit the role. As an example, a UiPath robot could send a resume, a job description, and a prompt to ChatGPT and request a numbered grade for the candidate. The grade returned by ChatGPT can be taken as an estimation of suitability for the job. Helping you speed up the task of reviewing all the resumes. ChatGPT becomes a ‘first line of review’, removing mismatched or poor-fitting candidates from the pool.
Benefit: Reduced time when filtering large volumes of resume submissions
To expand on the hiring theme, ChatGPT can also quickly generate effective interview questions based on job requirements and resumes. Candidate responses can also be evaluated and graded by ChatGPT if desired.
Hiring and interviews are a necessary but time-consuming activity for businesses. But with ChatGPT and UiPath robots, we can quickly build an automated process, connected into a recruitment platform, that creates a personalized set of questions for every candidate. Better interviews, more suitable hires, and in less time.
Benefit: Assists interviewer in preparing context-specific questions
If a company uses online chat for support discussions, ChatGPT can rank the quality of support for each case. For example, ask it to return a satisfaction score. If the score falls below a certain threshold, a robot could automatically escalate it to a manager for review. This will allow managers to enhance the training of their support teams. The evaluation of the response doesn’t have to be taken as absolutely correct, but it’s a good guide that can save precious time.
Benefit: Fast determination of support quality and the identification of escalations
Generative AI, like the kind found in ChatGPT, is going to be a valuable tool for developers of all levels. Its main use today will be when you need to generate something new (and non-critical!) based on certain specifications – or ‘prompts’ as they’re called in the new lingo. When you pair these results with action enabled by automation, you can tackle a broad set of interesting new use cases. What many organizations still need to work out is how to operationalize these generative AI tools in a business environment with consistency and governance.
Surely, this is just the start of the conversations on how you, me, and the rest of the world will use these tools. We plan on sharing more detailed ideas over the coming weeks and months. Register now to save your spot at the 2023 UiPath AI Summit.
A special thank you to Christian for his contributions to this article. To see him walk through these use cases and others, check him out on YouTube @CristianNegulescu. Also, a big thanks to Alvin Stanescu, Florin Grigoriu, Gerrit Knippschild, and other UiPathers for their contributions.
Editor's note: the views represented in this article are the author’s own and are not necessarily representative of (or an endorsement from) UiPath.
Sign up today and we'll email you the newest articles every week.
Thank you for subscribing! Each week, we'll send the best automation blog posts straight to your inbox.