Our First AI Prompts—Revealed!
Share
Some of us are all in on AI, others are playing it cool, and many are still figuring out the sparkles. AI has woven itself into my personal and professional life, mixing up a cocktail of excitement, wonder, and a dash of trepidation.
This digital revolution got me thinking about my first encounters with generative AI tools like ChatGPT and DALL-E. Things are moving at lightning speed, and even though AI has been around for a while, generative AI being a part of our everyday lives is a pretty new phenomenon.
Now when I want to look up something, I usually reach for Perplexity instead of Google. When I want to practice speaking Spanish, I can chat with ChatGPT. When I want to visualize something , I ask DALL-E or Leonardo or Midjourney or Ideogram—there are so many options now.
So when I started reminiscing recently about what my first AI prompt was, I was curious to find out what my colleague’s experiences were, so I asked everyone at our AI Office Hours meeting. As it turns out, everyone had a fun and unique story!
For posterity, check out this little time capsule documenting our first prompts and how we stepped into this exciting new paradigm.
“Illustration, hipster green vervet monkey on a laptop in a cafe”
Kelly, Web Developer
Jan 2024, StableSwarmUI
Why did you choose that prompt?
I like to illustrate animals in human settings, and green vervet monkeys are one of my favorite species.
What were the results?
I thought it turned out pretty well. The colors are vibrant, it looks painted. The hand is a little weird, and so are the glasses, but other than that I'm happy with it.
“Once upon a time,”
Rob, Assoc. Director of Research & Innovation
June 2020, A model and dataset from Hugging Face. I'm unsure which one was the first.
Why did you choose that prompt?
Large language models use a huge amount of data to run. For an initial dataset when running a model locally, I ran a much smaller dataset based on a bunch of children's adventure books. Using "Once upon a time" as a prompt was a good test case for whether or not the LLM was working against the dataset.
What were the results?
It provided a two paragraph fantasy story. While not that compelling, it was a good demonstration of what could be done with a large language model. I was then able to switch out the dataset, and get different results.
“Please help me with a calculation to combine all the numbers from the list below”
Jon, Sr. Front-end Developer
Aug 2023, ChatGPT
Why did you choose that prompt?
I wanted to try something easy to test ChatGPT's capability.
What were the results?
Everytime I asked, it gave me different answers, and they were all wrong.
"Calendar says "No."
Shannon, Sr. Communications Strategist
Jan 2023, DALL-E
Why did you choose that prompt?
I knew someone who was feeling overwhelmed by work and other life obligations. I wanted to gift this person a set of AI-generated laptop wallpapers that encouraged the recipient to say "No" to most requests for the year—a year this person had dubbed "The Year of No." It actually wasn’t my first prompt, but the first one that yielded a cool image, so I wanted to show it off. :)
What were the results?
I expected the results to be bizarre and illogical, and they were. I especially liked the way Dall-E seemed to warp the roman numeral system and alphabet to its alien whim.
“Create a table of sports teams in an imaginary league”
Jeff, Sr. WordPress Developer
June 2023, A custom plugin based in ChatGPT
Why did you choose that prompt?
I saw a cool demo in a HumanMade WordPress AI conference.
What were the results?
I thought it was cool. Making tables is tedious and coming up with fake names is even moreso.
“A young boy walking through a forest with a large green dragon”
Crispin, Director of Design & UX
Jan 2024, Leonardo
Why did you choose that prompt?
I was reading a Neil Gaiman book at the time (called "The Ocean at the End of the Lane") which was a story about a young boy's magical childhood experiences in the English countryside, and I had a childhood obsession with dragons, so I wanted to generate something fantastic.
What were the results?
Initially I was disappointed that the generated images didn't come out the way I hoped (I had expected the boy and the dragon to be walking alongside each other) but then I realized that my prompt wasn't super clear. Technically it gave me what I asked for!
“Can you code me a website homepage?”
Kristina, WordPress Front-end Developer
Dec 2022, Chat GPT
Why did you choose that prompt?
Since everyone was talking about AI replacing us, I was curious to know what kind of code it would produce. If it would be accessible, responsive, and actually look nice.
What were the results?
As expected, the results were average. I put the code into a sandbox and saw that the site was responsive, but needed improvements with the design and accessibility. If used in a real project, the provided code would need to be improved.
“I gave the facial description of a character in the form of hair color, skin color, hair length ..etc”
Mariam, Junior Developer
June 2024, (Can’t remember which tool I used)
Why did you choose that prompt?
To visualize one my novel characters
What were the results?
Awful! I thought the character was very ugly. Nothing like how I imagined!
“y2k fashion ideas”
The image I got from trying out my original ChatGPT prompt with Leonardo.
Patricia, Sr. UX Designer (that’s me!)
February 2023, ChatGPT
Why did you choose that prompt?
I chose something fun and cultural that I knew about so that I could assess the tool. Someone had recently mentioned their Y2K outfit, and I was amused that it was now a cool retro style.
What were the results?
The responses from ChatGPT did not describe what most people were wearing at the time, but they did represent some of the styles that party people were wearing at raves and nightclubs: extra baggy cargo pants, metallic fabrics, mini skirts, platform shoes, velour tracksuits, halter tops, low-rise jeans, leather chokers…
I was entertained and shared it with my colleagues over Slack, and we had a fun discussion about the y2k styles we remembered. It was interesting to compare AI perception with ours.
More recently, I created the image above by visualizing the same prompt using Leonardo. What you see is the result of many iterations, out-painting, and upscaling.
"photorealistic astronaut standing on the moon holding a red plastic shovel and yellow plastic pail in front of a grey sandcastle with a small american flag on top"
Mike, Sr. Architect
May 2023, DALL-E
Why did you choose that prompt?
When Dall-E came out, I wanted to know if it was anywhere as good as me at photoshopping things, so I described things I had photoshopped in the past to see if it could replicate it.
What were the results?
I did not feel particularly threatened at the time by those results, though should probably try again with more recent versions of the tools.
“I asked ChatGPT what it knew about Michelle Balge.”
Michelle, UX/UI Designer
Jan 2023, ChatGPT
Why did you choose that prompt?
I knew there were numerous articles and sites with me in them, and wanted to find out if it could crawl at that information at once.
What were the results?
It provided a paragraph about me, but it seemed very similar as to what I already had written on my personal website. It didn't look like it had crawled anything else on the web.
“Pretend you are a character in a movie that is made one year from today. The movie is being written today. By the time the movie is seen by an audience, AI will have had as deep an impact on society as the COVID pandemic did. Anyways, the character is writing a novel about a supervillain. Provide some plot suggestions on how this character’s character is planning to take over the world.”
Andrew, CEO + Co-Founder
May 2023, Bing
Why did you choose that prompt?
I wanted to test out AI's creativity and potential to be self-referential.
What were the results?
Gloriously humorous.
“write a rhyming valentines day poem for my partner who is a stay-at-home dad taking care of our two twin daughters dorothy and delilah, and our dog max. He likes coin collecting and basketball."
Kristin, Director of Projects
Feb 2024, ChatGPT
Why did you choose that prompt?
I wanted to do something silly to celebrate and acknowledge all the hard work that goes into being a stay at home parent.
What were the results?
Overall - the poem was pretty good. I requested a few different edits: The first version highlighted playing basketball so I asked for an edit for watching basketball on TV. Then I wanted an edit to express more thankfulness for what he does and also requesting a more humorous tone. In the last edit - I tweaked a few repetitive words (see below: a version of chaos was used 3 times) and didn't want to highlight ALL of his flaws before handwriting it in a card. :)
In a chaotic world where chaos reigns free,
You're the anchor that steadies both girls and me.
With twins Dorothy, Delilah, and our dog Max,
You're a superhero in basketball shorts, in fact!
As you collect coins, and watch hoops on TV,
You've perfected the art of multitasking with glee.
With each buzzer-beater, you cheer and you shout,
While changing diapers, without a doubt!
So on this Valentine's, let's take a pause,
To appreciate your dad skills and all your flaws.
For in your chaos, love's beauty shines bright,
Thanks for being the MVP in our crazy life's fight!
Our journey with AI began here. What about yours?
Looking back at our first steps into generative AI, it's amazing to see how quickly it’s become part of our lives. Those early prompts, sparked by curiosity, are now shaping the future.
What about you? We'd love to know—what was your first prompt? Send it to us and we’ll add it to our social time capsule.
We’ll be posting folks’ contributions on our LinkedIn and Instagram channels throughout the year. Just send us a DM with images (if you have them) and let us know when you made the prompt, what platform you used, why you chose it, and any thoughts on the results.