From Coding to Cooking: Practical Uses for ChatGPT

I described the friend I was buying for in some detail and mentioned a few ideas I had. ChatGPT mentioned a few other ideas and helped me come up with good options for an inscription.

I struggle to write cards. I care. I’m lousy at expressing it, though. I can give ChatGPT the basic sentiment I’m trying to express and a little about the recipient, and it gives me something that’s infinitely better than a blank card that I can customize to make my own.

Sometimes I need fake content to test site layouts or data processing. ChatGPT can create it for me, and even organize it in a comma-separated-value (CSV) format.

I know how to do the basics on a command line, but when I need a command that’s outside my norm, it’s almost always faster to explain to ChatGPT exactly what I want than to Google it. I get the precise command I need with no digging.

I had a funny idea for a novella and described it to ChatGPT, which turned it into a three-act outline, complete with characters, motivations, and outcomes. I asked it to write the opening scene in the style of some of my favorite authors, and while the results weren’t as good as their actual writing, they were definitely readable and on par with a lot of the cheap-but-fun sci-fi that abounds on Kindle Unlimited. If I had wanted to put the time and effort in, I’m confident I could collaborate with ChatGPT to write a passable book.

My teenage son is a mediocre Latin student, but once he realized he could talk with ChatGPT in Latin, he was suddenly practicing more simply for the sheer fun of it. This, of course, applies to most modern languages, too.

One of my tasks at work is to communicate complex technical processes in friendly language. A particular challenge I’ve had is coming up with shorthand terms to refer to different options. I wrote up a description of each option and asked ChatGPT to name them, and its suggestions were far better than what I’d come up with. My coworkers and our clients understand the options better as a result now.

The other day, I got a phone call from a scammer, so I wrote up my factual account to report them to the FTC. I also wanted to make sure the phone number they were using got listed on a site where it would be found by future googlers, so I asked ChatGPT to make my report more engaging. The resulting account was still accurate but written to be read by casual readers. I then submitted it to a few sites for that kind of thing and have every hope that it will spare someone the hassle of a talking to the scammer.

This one is from a friend of mine, who mentioned that she occasionally struggled to remain polite in her customer-facing support role. Most customers are lovely and communciation is straightforward, but sometimes a particular person is rude, demanding, or forgetful to the point where it’s hard to stay positive.

In those cases, my friend used to slog through writing her polite and professional email responses. Now she quickly fires off a candid reply—to ChatGPT. And then she asks it to rewrite it to be polite and professional, and that’s what the customer receives. It’s the same result for the customer, but with far less agonizing over the right phrase for my friend.

I recently learned about plain language and Easy Read and was curious if ChatGPT could help produce this accessible content. It can. 🙂

This is super useful as a starting point. You can ask ChatGPT for a job description for pretty much anything and then just tweak it to meet your specific needs.

After we got good-but-not-great candidates for an open position, we reflected on what qualities were missing. We then gave ChatGPT our job posting and told it our recruiting struggles. It refined the posting to emphasize the qualities we needed. The next round of applicants were better.

In passing, an acquaintence mentioned a well-known authority on ADHD, and I was curious about their distinctive perspectives. I asked ChatGPT and it gave me bullet points about the most relevant positions while also noting where other ADHD authorities disagree or have differing views.

I often think of questions like “What’s the difference between a data analyst and a data scientist?” They aren’t important enough to read a whole article about, simply a passing curiosity. With ChatGPT, I ask and get a quick answer and move on with my day.

I was reading an older book that mentions “the Christian Occidental view of man.” I wasn’t familiar with the term “Occidental,” let alone the broader concept. ChatGPT explained both.

I was reading an organization’s description of their environmental stewardship policy, and it smacked of greenwashing but I couldn’t put my finger on what exactly was wrong with it. I asked ChatGPT to evaluate it and it helped me understand exactly what was snagging my brain.

The thing I love about this use is that we can use it proactively for our own work. ChatGPT can act as a disinterested third-party to poke holes in arguments and help us see our own blind spots, and improve not just our communication but our policies themselves as a result.

This is one of the tasks at which ChatGPT excels. I had to laugh when I sent a coworker an article about innovative ways that some educators are using ChatGPT, and he promptly used ChatGPT to summarize it. What I thought was interesting, though, was his reasoning: he wanted to quickly understand it so we could continue our conversation. When I think of all the conversations I “mean to get back to,” this seems particularly smart.

YouTube videos frequently have transcripts, and they can also be copy/pasted into ChatGPT for a summary or highlights. I’ve sometimes run into length issues (ChatGPT will only take a certain amount of input) but you can split it if needed.

One of my coworkers recently did a video training and once he added it to Vimeo, it created a transcript. He gave that to ChatGPT, lightly edited the main points it gave him back, and put them into Slite, our knowledge management tool. Slite has its own AI implementation, so having that information in text form greatly improves the answers that Slite will give us to future questions.

I read a couple of the StrengthsFinder books years ago and got my list of strengths. I don’t really remember the specifics anymore, though, and realized that I’d get more out of the information if I had it in a quick-reference form. Getting ChatGPT to give me a summary, application, and drawbacks for each of my strengths has put this information at my fingertips.

Due to childhood events, I have awkward relationships with some of my extended family. However, I still choose to see them occasionally. ChatGPT was incredibly helpful recently at suggesting activities we can do together that fit our varied personalities and give us something to focus on other than the awkward silence.

I wrote up detailed lists of characteristics of a couple of teenagers I know (with their help) and answers to some of the kind of standard career questions you come across. ChatGPT did a great job of providing a list of potential careers and tying them back to the relevant characteristics.

A few friends and I fantasize about having a co-op or commune or something when we retire. Over the years, we’ve made a pretty extensive (and diverse) list of preferences for our retirement location—everything from solar index to distance from a Southwest-served airport to the social and thermal acceptibility of wearing flannel. I gave ChatGPT our list of criteria and it suggested a number of locations that meet them. Several of them were places we’re already considering, and some were new ideas but good ones.

A relatively new friend described herself as an Enneagram 3. I felt like this was a very helpful tidbit to learn a lot about how she sees herself and the world. ChatGPT gave me a list of common associations and also a number of on-point suggestions about how to support my friend and bring out the best in her. I kind of want this for all of my friends now.

I wanted book recommendations on a fairly specific topic (reconciling the depictions of the Biblical God in the Old Testament with Jesus in the New Testament) and from authors with a particular theological perspective. I gave ChatGPT my request with a few sample authors, and it provided a number of matching books. I then asked it to format the results with links to Goodreads (they have a predictable structure for their search URLs) and all of them were actual books that were on topic. (Now they’re on my ridiculous to-read list.)

One of my coworkers joked that our company should buy us all Steam Decks for socializing purposes (we’re fully remote) and as a joke, I asked ChatGPT to write an argument in favor of the benefit. It was a bit over the top, but if I wanted to write a real case, I’d definitely start with a ChatGPT draft and refine from there. (No Steam Decks yet, BTW.)

ChatGPT will be the first to tell you that it can’t provide legal advice and that you should check with a professional, which, you know, is true. I had a low-stakes question (about whether a teenager volunteering requires a work permit in California) and it gave me what seems to be an accurate answer. Your mileage may vary!

I use this one a lot. Many of my kitchen experiments leave me with odds and ends (like the pulp from making oat milk or carrot juice) and sometimes I buy an interesting-looking ingredient without a plan. I’ve had surprising success asking ChatGPT to make general suggestions for an ingredient and having it create a recipe for me to follow. My husband was very skeptical initially, but all of the recipes I’ve made have been great.

I often adjust recipes, cook freestyle, or attempt to optimize kitchen processes. I’ve been delighted with ChatGPT’s feedback on questions like “what internal temperature does an egg-based casserole need to reach to set?” and “does pre-hydrating psyllium powder affect its nutritional benefits?” (among many others).

I like fancy food and I also like quick-and-easy food that comes out of boxes and packets. I had some ramen noodles that don’t come with seasoning packets (they’re an ingredient, not a Top Ramen–style product). ChatGPT was able to give me a recipe for a basic seasoning mix using predictable ingredients like bouillon powder, but also variations using whole-food ingredients and less-common flavors.

I was making homemade almond milk based on a questionable recipe (it didn’t include salt). I like store-bought almond milk, so I looked at the nutrition facts, and then asked ChatGPT extrapolate the amount of table salt I should add based on the sodium number per serving and the amount of servings I was making. (Alas, it turns out I don’t like homemade almond milk, but not because of the salt amount!)

I am a bit of a dietary weirdo in that I prioritize some preferences for nutritional or social reasons (like whole foods and Fair Trade) but don’t necessarily care about the preferences that usually go along with them (most Fair Trade products are organic; many whole food–centric recipes are vegan). ChatGPT has been very helpful in working up recipes that combine my preferences in unique ways (like a cashew-based mayonnaise that uses egg yolks).

I remember some delicious, delicious sprouted wheat bread from my childhood (shoutout to Price Chopper deli sandwiches!) and had some wheat berries, so I got the step-by-step process for sprouting them and then using them in my bread recipe. It also helped me through the process of dehydrating the sprouted wheat berries and grinding them into flour.

I’m an aspiring patio gardener. My patio is a touch challenging, though—it’s north-facing, tiny, and surrounded by a solid six-foot fence. I’ve gotten great advice from ChatGPT about plant selection and care based on my growing zone, patio specifics, constraints (containers only!), and preferences (I like edible plants).

This is one of my favorite kinds of requests for ChatGPT: a scenario where there’s vast amounts of data available online, but it’s hard to find detailed information for your exact situation. ChatGPT’s ability to synthesize and customize makes it exceptionally helpful for this kind of query.

This is kind of an extension of the previous use. I’ve gotten very interested in food forests, and while I’m not in a position to do anything like that now, I’m dreaming for the future. ChatGPT’s training familiarized it with the concept of food forests, which plants fall into which categories, and which can grow in which hardiness zones. It combines all of those to provide a detailed list of plants for this method of food-growing.

In southern California where I live, in spring the hills turn neon yellow with a flower that I learned is black mustard. It’s an invasive species and the bane of local ecologists, but it’s also completely edible. Since it’s already here, I was curious to try making mustard from its flowers, but also loathe to do anything that would spread it further. With ChatGPT’s help, I learned about the plant’s life cycle and when it’s safe to harvest vs. going to spread the problem.

I’m well-versed in some programming languages, but have little experience in others. When I was following a tutorial for a variant of JavaScript that I’m less familar with, I wondered about the purpose of some syntax. This is the kind of thing that’s been frustrating to Google in the past, because it’s pure punctuation (which Google discards) and I don’t know the name of the syntax yet.

Giving the chunk of code to ChatGPT and asking about that specific syntax was a joy. It explained what the syntax did in general terms, what it was doing in that particular section of code, and when to use it versus some other approach. Instead of merely “getting the gist from context” like I used to, I am learning new and useful information constantly in ways that lead to application.

I care deeply about well-commented code, but don’t love writing general information comments like DocBlocks. Fortunately, when I feed ChatGPT a function, it can give me back a DocBlock in a snap.

Note that GitHub Copilot is also excellent for this, and is even more convenient because it’s running in my code editor. But GitHub Copilot comes with a monthly fee, which might be off-putting.

I’ve used the WordPress Coding Standards for long enough that I follow them without thinking when I’m writing code. However, when I’m copying code from another source, or when ChatGPT is generating code, it often follows another standard (or none at all, in the case of some Stack Overflow posts). ChatGPT can easily rewrite code to match WPCS (or any other well-documented standard), saving me tedious updating.

Over time, best practices change. I had an old class that used now-deprecated syntax for creating arrays in PHP. ChatGPT can take the class as an input and give me back a PHP 8-friendly version without changing anything else.

Similarly, I wrote some SQL that used an old style of joining tables. ChatGPT gently informed me that it made more sense to use a LEFT JOIN instead and gave me back the improved version.

This applies to code but also pretty much everything else. If you give ChatGPT a chunk of code or writing and identify it as something you’re working on, it will typically summarize what you have (usually with a compliment!) and provide suggestions for improvement. I like this a lot because I don’t even need to have a hunch about what might be weak before getting feedback.

Sometimes I have an idea of what I want to accomplish and know there’s a function for it but don’t remember if it’s a native PHP function or a WordPress function, and don’t remember the exact name. In these cases, I can tell ChatGPT what my goal is (things like “I want to sanitize a user-inputted value but I need to keep the punctuation intact”) and what language/platform I’m using. Then it will tell me which function suits my needs.

ChatGPT tends to do a great job on generic programming tasks using common languages. It understandably struggles a bit more with specific APIs and less-popular tools and frameworks. In those cases, I feed it relevant documentation and treat it as a pair programmer with similar skills to my own. I question its output, report back errors, make suggestions, and give it updated versions of the code that I’ve adjusted. Thie still saves me time versus starting from scratch, and it’s making me a better programmer to boot.

I’ve been surprised and impressed at how quickly and easily ChatGPT has been able to write complex queries in niche syntax when I give it the latest documentation. Once it knows the right structure, I’m able to describe plain-English goals and it spits out the correct query for me to use.

A lot of programming error messages are cryptic, and googling them leads down a lot of false trails because they often happen in situations that have almost nothing in common. With ChatGPT, it either already knows the context in which the error is happening (if it happens along the way in a larger conversation) or I can tell it the context and it will take that into account. This almost always leads to valid answers much faster than googling.

We had a client ask for a very specific way of interacting with the taxonomy on their WordPress editor pages. It made sense for them, but it wasn’t the kind of thing that would be broadly applicable to any of the rest of our clients. We went searching for an existing plugin, and found one, but it was limited to two levels of hierarchical terms, and our client had three or four levels.

I identified the relevant function, told ChatGPT that we wanted to update it to handle any number of hierachical levels, and gave it the function. It returned a better-written version with the necessary updates to make it work recursively, and we’re off to the races.

I came across a reference to the “Behavior-driven development (BDD)” software methodology and wanted to quickly understand more about it. ChatGPT gave me the key concepts, and then I gave it a one-sentence overview of a type of work I do and asked for concrete examples. It gave me some excellent use cases and the conversation then wandered into how to implement it practically in my workflow and tech stack.

Knowing both tools help me write code, I asked for a comparison and ChatGPT gave me specifics of each tools’ strengths and suggestions for when to use which tool. “ChatGPT for general programming advice, high-level design discussions, and non-programming related topics, while using GitHub Copilot for context-aware code generation, syntax/API assistance, and refactoring.”

I wanted to connect the Asana API with the Motion API to sync tasks between the two systems automatically. I’ve never really used Python but ChatGPT loves it, so I figured it was a good opportunity to learn a little bit through doing (and I mean letting ChatGPT do it!). Whenever I didn’t understand what was happening, I’d ask for an explanation and I learned a lot of general principles (and got my integration working well).

When I’m working with different APIs, I test things out in Postman before creating the relevant code. One API I was playing with needed my request to be formatted differently than I usually do, and I wasn’t sure what the correct syntax was when I wanted to adjust their example. I fed the example to ChatGPT, explained what I wanted to change, and got back a body structure that worked perfectly.

I use Obsidian as my note-taking tool of choice and I was having some trouble getting it set up with Git LFS. The error messages weren’t particularly helpful, and I wasn’t sure which part of the process (Obisidan, Obsidian Git, GitHub, or Git LFS) was the problem. I fed ChatGPT context and error messages and got it sorted out beautifully.

I’m using Algolia as the index for a work project but after getting everything up and running, discovered a bug that was creating some junk in my records. I used ChatGPT to write a script to update the broken records and give the client a quick fix while I dealt with the underlying bug.

I do a lot of migrations from old databases and CMSs, and I’ve been loving how I can quickly give ChatGPT a description of the tables and an explanation of how I’d like the data connected and returned. It figures out which kind of JOINs to use and gets me exactly what I need.

I had a lot of records that had member and non-member prices entered in an open-ended text field and they were wildly inconsistent. I gave ChatGPT a representative sample of all the different ways prices were represented and got a regex pattern to help me sort out the classified prices and get them into two separate lists.

I use Google Sheets conditional formatting fairly rarely and always have to look up examples on how to do row-level highlighting based on the value of a column. With ChatGPT, I just explained which column and conditions should be considered, and it gave me the custom formula to paste in.

I have a random idea for some information formatted as trading cards, and I gave ChatGPT a list of the information they’d contain. It created the HTML and CSS to lay out the cards for printing purposes, with placeholders for the actual data.