photo_Pawel/Getty Images
One of the more intriguing discoveries about ChatGPT is that it can write pretty good code. I first tested this out in 2023 when I asked it to write a WordPress plugin my wife could use on her website. ChatGPT did a fine job, but it was a simple project.
Also: The best AI for coding in 2025 (and what not to use)
So, how can you use ChatGPT to write code as part of your daily coding practice? Here's a quick summary:
ChatGPT can produce both useful and unusable code. For best results, provide clear and detailed prompts. ChatGPT excels in assisting with specific coding tasks or routines, rather than building complete applications from scratch. Use ChatGPT to find and choose the right coding libraries for specific purposes, and engage in an interactive discussion to narrow your options. Be cautious about who owns AI-generated code and always verify the code's reliability. Don't blindly trust the generated output. Treat interactions with ChatGPT as a conversation. Refine your questions based on the AI's responses to get closer to the desired output. Paid-tier users gain access to some additional features for more comprehensive coding support, including Deep Research analysis and Codex full installation coding.
Now, let's explore ChatGPT in considerably more depth.
What types of coding can ChatGPT do well?
There are two important facts about ChatGPT and coding. First, the AI can write useful code.
The second is that the AI can get completely lost, fall into a rabbit hole, chase its tail, and produce unusable garbage.
Also: Tech prophet Mary Meeker just dropped a massive report on AI trends - here's your TL;DR
I found this fact out the hard way. After I finished the WordPress plugin for my wife, I decided to see how far ChatGPT could go.
I wrote a very careful prompt for a Mac application, including detailed descriptions of user interface elements, interactions, what would be provided in settings, how they would work, and more. Then I fed the prompt to ChatGPT.
ChatGPT responded with a flood of text and code. Then it stopped mid-code. When I asked the AI to continue, it vomited even more code and text. I requested continue after continue, and it dumped out more and more code. However, none of the output was usable. The AI didn't identify where the code should go, how to construct the project, and -- when I looked carefully at the code produced -- it left out major operations I requested, leaving in simple text descriptions stating "program logic goes here."
Also: How ChatGPT scanned 170k lines of code in seconds and saved me hours of work
After repeated tests, it became clear that if you ask ChatGPT to deliver a complete application, the tool will fail. A corollary to this observation is that if you know nothing about coding and want ChatGPT to build something, it will fail.
Where ChatGPT succeeds -- and does so very well -- is in helping someone who already knows how to code to build specific routines and get tasks done. Don't ask for an app that runs on the menu bar. But if you ask ChatGPT for a routine to put a menu on the menu bar and paste that into your project, the tool will do quite well. The new Codex feature for paid tiers might expand on this, but even OpenAI says that Codex is best for maintenance and support rather than doing all original writing.
Also, remember that while ChatGPT appears to have a tremendous amount of domain-specific knowledge (and often does), it lacks wisdom. As such, the tool may be able to write code, but it won't be able to write code containing the nuances for specific or complex problems that require deep experience.
I have worked with ChatGPT a lot while trying to write and debug code, and some of the inane suggestions the AI has made have been stunningly stupid and unhelpful. But it has also proved helpful, which is why I keep using it. You just need to be able to differentiate between the two.
Also: How AI coding agents could destroy open source software
Use ChatGPT to demo techniques, write small algorithms, and produce subroutines. You can even get ChatGPT to help you break down a bigger project into chunks and then you can ask it to help you code those chunks.
With that in mind, let's look at some specific steps for how ChatGPT can help you write code.
How to use ChatGPT to write code
1. Narrow down and sharpen up your request This first step is to decide what you will ask of ChatGPT -- but not yet ask it anything. Decide what you want your function or routine to do, or what you want to learn to incorporate into your code. Decide on the parameters you'll pass into your code and what you want to get out. And then look at how you're going to describe it. Also: How to write better ChatGPT prompts Imagine you're paying a human programmer to do this task. Are you giving that person enough information to be able to work on your assignment? Or are you too vague and the person you're paying is more likely to ask questions or turn in something entirely unrelated to what you want? Here's an example. Let's say I want to be able to summarize any web page. I want to feed the AI this article and get back a well-considered and appropriate summary. As my input, I'll specify a web page URL. As my output, it's a block of text with a summary. Show more
2. Use ChatGPT to explore libraries and resources Continuing with the example above, an old school way of extracting web page data was to find the text between HTML paragraph tags. However, with the rise of AI tools, you can use an AI library to do an intelligent extract and summary. One of the places ChatGPT excels (and it's also an area you can easily verify to avoid its authoritative-but-wrong behavior pattern) is finding libraries and resources. Also: The best free AI courses OpenAI (the maker of ChatGPT) sells API access to its LLMs to do exactly what we want. But in the case of this example, let's assume we don't want to pay transaction fees. So, let's look at interacting with ChatGPT to figure out how to use such a tool, for free, with a project that runs in PHP. Show more
I started with a prompt to elicit information about what libraries would provide the desired functionality. A library (for those reading along who aren't programmers) is a body of code a programmer can access that does a lot of the heavy lifting for a specific purpose. A big part of modern programming is finding and choosing the right libraries, so this is a good starting point.
In this case, I'm looking at blocks of code written by other people that will summarize text. Here's my first prompt:
Describe ten different open source AI libraries (and the languages they work with) that I can use to generate a summary of the main core contents of any web page, ignoring any ads or embedded materials.
This prompt gave me exactly what I wanted, including a mention of OpenAI's offerings. I think OpenAI would do great here, but for this hypothetical project, I don't want to budget for API fees. So I'll narrow down the question:
Are any of these free?
ChatGPT hedged its bets with its answer: "Yes, all ten of these AI libraries are open source and free to use. However, some of them may have usage limits or require payment for access to additional features or resources." So, based on that response, I clarified my query:
Which of these libraries have no usage limits and don't require any additional payment or licensing?
Notice how this is very much a conversation. I don't have to re-ask the original question -- I'm just drilling down as I might if I had an expert next to me. This time, ChatGPT gave me eight library choices, but none mentioned the PHP language I was planning to use to code. So here's the next prompt:
Of those eight libraries, can I use any with PHP?
It returned three libraries, but I wasn't sure what each did. So I asked another question:
What's the difference between Sumy, Gensim, and NLTK?
I still wasn't sure, so I clarified my use plan and then asked:
If I want to create summaries of web-page news articles, which library would work better?
The answer was clear and promising: "Sumy is specifically designed for text summarization, which is the task of creating a summary that captures the most important information from a piece of text." So it was time to see what was involved in using Sumy with PHP. I asked my last question for this part of the project:
Can you explain how to use Sumy from PHP?
Feel free to play along on your computer and paste these prompts into ChatGPT. Notice that, in step one, I decided what program module to get help on. Then, in this step, I had a conversation with ChatGPT to decide what library to use and how to integrate it into my project.
Also: The best AI chatbots
That approach might not seem like programming, but I assure you it is. Programming isn't just blasting lines of code onto a page. Programming is figuring out how to integrate all the various resources and systems, and how to talk to all the components of your solution. Here, ChatGPT helped me do that integration analysis.
By the way, I was curious whether Google's Gemini AI could help similarly. Gemini did give some extra insights into the planning aspect of programming over ChatGPT's responses.
So don't hesitate to use multiple tools to triangulate your answers. Both Google Gemini and Microsoft Copilot have improved radically in the last few years. Early on, they (to be polite) sucked. Now they're really pretty darned good. Gemini has Jules, which competes against ChatGPT's Codex. I haven't yet been able to test Codex, but Jules is impressive as heck. These are so-called "agentic" tools which run as agents behind the scenes, usually modifying GitHub repositories.
Also: How I test an AI chatbot's coding ability - and you can too
Coding is next.
3. Ask ChatGPT to write example code OK, let's pause here. This article is entitled "How to use ChatGPT to write code." And it will. But what we're really doing is asking ChatGPT to write example code. Also: The rise and fall in programming languages' popularity since 2016 - and what it tells us Here, things have changed over the last few months. I previously wrote this: Unless you're writing a small function (like the line sorter/randomizer ChatGPT wrote for my wife), ChatGPT can't write your final code. First, you'll have to maintain it. ChatGPT is terrible at modifying already-written code. Terrible, as in, it doesn't do it. So, to get fresh code, you have to ask ChatGPT to generate something new. As I found previously, even if your prompt is virtually identical, ChatGPT may unexpectedly change what it gives you. This is still the case for basic ChatGPT: ChatGPT can't maintain your code, or even tweak it. Show more
That limitation means you have to do the legwork yourself. As we know, the first draft of a piece of code is rarely the final code. So, even if you expect ChatGPT to generate final code, it would be a starting point, and one where you need to take it to completion, integrate it into your bigger project, test it, refine it, debug it, and so on.
But now there's Codex, and its entire purpose is maintaining already-existing code. Codex has a lot of limitations, including not remembering anything between individual instructions. It also lives in a completely different context from the rest of your ChatGPT interactions, so it's almost like you're using two completely different, unrelated tools. But Codex can maintain. So that's new.
But even if you don't use Codex, that doesn't mean the example code is worthless -- far from it. Let's look at a prompt I wrote based on the project I described earlier. Here's the first part:
Write a PHP function called summarize_article. As input, summarize_article will be passed a URL to an article on a news-related site like ZDNET.com or Reuters.com.
I'm telling ChatGPT the programming language it should use. I'm also telling the AI the input and providing two sites as samples to help ChatGPT understand the article style. Honestly, I'm not sure ChatGPT didn't ignore that bit of guidance. Next, I'll tell it how to do the bulk of the work:
Inside summarize_article, retrieve the contents of the web page at the URL provided. Using the library Sumy from within PHP and any other libraries necessary, extract the main body of the article, ignoring any ads or embedded materials, and summarize it to approximately 50 words. Make sure the summary consists of complete sentences. You can go above the 50 words to finish the last sentence, if necessary.
This approach is very similar to how I'd instruct an employee. I'd want that person to know that they weren't only restricted to Sumy. If they needed another tool, I wanted them to use it.
Also: IBM will train you in AI fundamentals for free, and give you a skill credential - in 10 hours
I also specified an approximate number of words to create bounds for what I wanted as a summary. A later version of the routine might take that number as a parameter. I then ended by saying what I wanted as a result:
Once processing is complete, code summarize_article so it returns the summary in plain text.
The resulting code is pretty simple. ChatGPT called on another library (Goose) to retrieve the article contents. It then passed that summary to Sumy with a 50-word limit and returned the result. But once the basics are written, it's a mere matter of programming to go back in and add tweaks, customize what's passed to the two libraries, and deliver the results:
Screenshot by David Gewirtz/ZDNET
One interesting point of note: When I originally tried this test in early 2023, ChatGPT created a sample call to the routine it wrote, using a URL from after 2021. At that time, in March 2023, ChatGPT's dataset only went to 2021. Now, the ChatGPT knowledge base extends to the end of June 2024 and can search the web. But my point is that ChatGPT made up a sample link that it couldn't possibly know about:
https://www.reuters.com/business/retail-consumer/teslas-musk-says-fremont-california-factory-may-be-sold-chip-shortage-bites-2022-03-18/
I checked that URL against Reuters' site and the Wayback Machine, and it doesn't exist. Never assume ChatGPT is accurate. Always double-check everything it gives you.
4. Debug and refine the generated code I showed you a few ways that ChatGPT makes mistakes or hallucinates. All programmers make mistakes, even the AI ones. But you can do several things to help refine your code, debug problems, and anticipate errors that might crop up. My favorite new AI-enabled trick is to feed code to a different ChatGPT session (or a different chatbot entirely) and ask, "What's wrong with this code?" Show more