2024, June 5

hey chatgpt, on a scale from 0 to 10, where 0 is super complicated and clunky, and 10 is totally smooth, simple, and flowing, where would you rate your own phrasing right now?

when i turn to chatgpt for phrasing suggestions, its standard responses often feel boring and useless. so i’ve started challenging the ai to scale its own text suggestions. it’s interesting to see how engaging in this self-feedback mode shakes things up.

i love to play around with a wide range of different scales, testing various points to see what happens. sometimes, just to see what occurs, i ask for a response at scale point 0 and then contrast it with one at scale point 10. sometimes, i crank it up by requesting a response at scale point 15 on a 0-10 scale.

more examples of scales for experimenting with chatgpt:

• translation scale: from 0, a literal and direct translation, to 10, where it becomes much freer and echoes how a native speaker would naturally express it.

• boldness scale: from 0, very polite and cautious, to 10, rebellious, cheeky, and audacious.

• passivity to activity scale: from 0, completely passive and unengaged, to 10, highly active and proactive.

• calm to explosive scale: from 0, very calm and subdued, to 10, super explosive and on the verge of freaking out.

• conventionality scale: from 0, being highly predictable and conventional, likely to be accepted by a mainstream audience, to 10, being so unique and unconventional that it’s unlikely to be embraced by a mainstream audience. in applying this scale, i ask chatgpt to draw from all previous inputs i’ve given and select those ideas that would be least accepted by the mainstream, pushing the boundaries of conventionality even further.

give it a try and have fun experimenting!

MORE WRITING PROMPTS

2024, may 26

when brainstorming, writing, and editing texts with chatgpt, i often find its standard responses clichéd, generic, and annoying. so i switched to giving chatgpt personas, like a kafkaesque beetle, to get more refreshing responses. turns out, it works a million times better than yelling at chatgpt to stop being such a cliché machine. these persona switches have been a huge game changer in my creative work with chatgpt.

it does take some experimentation, though. not every persona will give satisfying responses right away. it’s worth playing around with different personas, intensifying their edginess or craziness, or tweaking them to suit your needs. the key is to keep experimenting and being playful to see what new insights emerge.

examples of persona prompts:

• pumuckl:

(for those not familiar, pumuckl is a mischievous, invisible goblin from a german children’s book, known for his playful and chaotic nature.)

hey, chatgpt, imagine you’re pumuckl. i’m sending you several pages from my morning diary, and i want you to use your pumuckl mind to extract the most intriguing and surprising details, descriptions, and unconventional images. focus on those elements that break expectations and would never appear in a mainstream magazine. list these out so we can shape them into a new text. remember, you are pumuckl.

• subversive feminist new york times editor from the future

hey, chatgpt, imagine you’re a radical, subversive, feminist new york times editor living in the 2070s, looking back at the 2020s. you are intrigued by the zeitgeist of the 2020s and the unique quirks of different people’s lives. i’ll share a period from my recent life that i’d like to turn into essayistic, autobiographical chapters. let’s brainstorm how we could structure these into chapters, and i’d love your ideas on chapter titles and key themes. also, let me know which aspects intrigue you the most.

• kafkaesque beetle

hey, chatgpt, imagine you’re a kafkaesque beetle. i’m going to share some events from the life of my main character in my novel. describe these events from your perspective as a kafkaesque beetle.

MORE WRITING PROMPTS

2024, april 17

voice-activated chatgpt really has revolutionized my writing process. i now get most of my writing done by talking to chatgpt while i’m walking. it’s been the biggest game-changer of my creative life.

it’s like having a phone call with chatgpt while i’m walking. two things happen on that call: 1. as i speak, my thoughts are being transcribed. huge chunks of my writing get done here because this already provides a written record of my ideas. 2. chatgpt then reflects these ideas back to me, which works as a kind of reshaping of my words for me. sometimes i like the sentences chatgpt provides, and it helps me clarify what i want to express. sometimes i hugely dislike chatgpt’s words and ideas, and this forces me to shape my ideas with even greater distinction.

(good to know: i have chatgpt‘s paid version, and i’m using it as an app on my phone — i’m not sure if voice activation comes with the free version. if not, i definitely recommend trying out the paid version for a month; last time i checked, you could cancel your subscription monthly.)

here are the main prompts that i use for my walk ’n write sessions with chatgpt:

prompts for walk ’n write with chatgpt

• kick off: “hi chatgpt, i’m about to write an essay. i want to share all the thoughts that pop up in my mind that might end up in my essay. are you ready?”

• mid-conversation: sharing 1-3 thoughts per output. it should be fast and short, otherwise, i will overwhelm chatgpt. i close the output with “reflect back to me what you just heard me saying, without overly praising me. prioritize the thoughts and sentences that you find most edgy and surprising and most unlikely to be accepted by some kind of mainstream magazine.”

• end summary: “list 9 key points from our conversation that seem the most significant for my essay. prioritize the key points that you find most edgy and surprising and most unlikely to be accepted by a mainstream magazine. for each key point, provide a mini-summary.”

• after my walk ’n write session with chatgpt i usually go home to transform the transcript into a coherent piece of writing.

why no praising? in my experience, chatgpt tends to overpraise, which clutters the actual creative process i want to engage in.

why anti-mainstream? in my experience, chatgpt tends to ignore the most unusual parts and overprioritizes generic thoughts and clichés. even if my original thought is unconventional, chatgpt will turn it into the most generic cliché thing ever. that’s why it needs prompts that instruct it to choose the things that are unlikely to be accepted by the mainstream.

there are a thousand other playful ways to trick chatgpt into giving more interesting, unconventional responses. stay tuned for more.

MORE WRITING PROMPTS