Contact us
5 minutes read

One of the most intriguing things about Chat GPT is that it can write code.

Actually, there are two vital things to mind about Chat GPT and coding. First, it’s capable of writing useful code. The second is that it can write completely useless garbage. 

But how powerful is Chat GPT? Is it possible to use it to generate code for any type of programming language, from the most basic to more complex systems?

Our developer decided to use it for a range of purposes during their work on the project, and here are some tips we concluded after running a bunch of tests:

Narrow Down and Clarify Your Request

Decide what you want to ask Chat GPT, what parameters you'll pass in, and how you'll describe it. Ensure you provide enough information for Chat GPT to understand your request.

The following case illustrates the situation:

Chat GPT
In this example, we explain the task without going into detail about context, programming language, and description of the final goal.
Chat GPT
Chat GPT

In the second example, we give Chat GPT more context on what exactly we need to do, and what is the programming language and framework. Though like in the first example, we failed to indicate that the job should work at 3 AM EST.

As a result, AI successfully described the problem but didn’t manage to completely solve the problem we needed.

Chat GPT

Finally, we add the message to our chat that the job should work at 3 AM EST. 

Chat GPT analyzes it, compiles it with that answer, and corrects it with the example that this framework ( boilerplate) has such a feature, and you can set cron in connected jobs.

Chat GPT

Next, we ask Chat GPT to cover everything with tests, again, specifying the context, and what framework we need to cover with tests. For now, we are not going to code, only when we have the implementation of some functionality and tests covered for that.

Working with the code, we understand that we can achieve better results by using an asynchronous method in this case.

Since we are already in the same context, we ask Chat GPT to change to an asynchronous one. (Here it is important that the GPT should be aware that when it was learning and analyzing this lib, at that time there was this asynchronous code in the framework).

But in this case, it didn't generate the needed code, it just blindly made an asynchronous method that doesn't exist in that implementation.

Therefore, we decided to start a new chat and set the correct connection from the beginning.

Chat GPT
Here, we already provide a direct context of the expected results and even the class to be implemented.
Chat GPT
The Chat decides itself that the job will work in Quartz, and not in Hangfire (which we use). Though it couldn’t know it as we didn’t provide this information in the chat.
Chat GPT

Use It Instead of Google

For a long time, Google has been a go-to search engine for coding help. However, Chat GPT is quickly taking its place, as it offers a more interactive and personalized experience, “talking” directly to you and answering your specific questions.

In the following case, business analytics didn’t specify the requirements for the phone number feature. Instead of googling the specific requirements of minimum/maximum lengths of phone numbers for each country, we simply asked Chat GPT:

Chat GPT use case

The results were later checked and approved by the QA team.

Solve Complex Calculations to Save Time

Apart from its other capacities, Chat GPT stands out due to its remarkable mathematical capabilities. Its extensive training data and in-depth knowledge of complex mathematical concepts enable it to answer questions beyond basic math queries.

In the following example, we needed to calculate the difference between two dates in minutes. Instead of writing a console application or using online services, we just gave Chat GPT the task to calculate and in a matter of seconds, received the results.

Chat GPT use case

The calculation illustrated is quite simple though Chat GPT can be used to solve more complex calculations and equations. Just provide correct commands and parameters, and get an accurate result in a few seconds.

Brainstorm with Chat GPT

Chat GPT use case

Instead of reading various resources on the web, we just asked about best practices to track web devices for two-factor authentication, and Chat GPT gave us a summary of the most popular options. Of course, the answer can’t be taken for granted and must be analyzed in correlation with your specific case. Though this way, devs can save up time and concentrate on going in-depth only with the options that suit their project.

Key Takeaways

From the analysis we carried out, we can drive the following conclusions:

  • If you ask Chat GPT to write a complete application, it fails. The same thing happens if you know nothing of coding and hope that it will build high-quality code for you.

  • Where Chat GPT succeeds is in helping an experienced developer accomplish specific tasks and build specific routines. 

  • The results it gives should be analyzed and adjusted to your specific case.
7 people like this

This website uses cookies to ensure you get the best experience on our website.

Learn more
Thank you for getting in touch!
We'll get back to you soon.
Sending error!
Please try again later.
Thank you, your message has been sent.
Please try again later, or contact directly through email:
Format: doc, docx, rtf, txt, odt, pdf (5Mb max size)
Thank you, your message has been sent.
Please try again later, or contact directly through email: