{D} Not caring, but harder
Vol. 6, Issue 179 // pics + links // maybe try clicking the pics too
Hey there from the high valley of Mexico City.
I’ve started to receive, more and more often, feedback from humans that is written by AI.
First frustrating thing: sometimes, this feedback is good, lol!
Second frustrating thing: more often, this feedback is poor because it lacks context, which makes the revision process take longer.
Third (and most) frustrating thing: sometimes, the person giving the feedback partially re-writes the copy to not sound like an LLM wrote it — but tell-tale signs remain.
People: please don’t do this.
Not just because it’s kinda insulting, but also because it actively slows down the work and introduces a whole caboodle of doubt.
Let me show you what I mean using the example of a current workflow.
Right now I’m writing scripts for a video series. There are three phases to script writing: research, outline draft, script draft.
Research is done with a custom skill in Claude (great LLM use case: get up to speed quickly). Outlining is done with a custom skill in Claude (great use case: use LLMs to organize information).
Writing I do myself — not because I’m a great writer, but because Claude is a very bad writer.
Even with significant prompting and guardrails, Claude returns awkward copy. It uses the same triadics, and the same pacing. It adds in superfluous details. It uses a mediocre voice.
It’s much easier to achieve a higher bar of writing quality when you limit Claude to organizing data and use your own brain to express ideas using that organized data. I know because I have tried to get Claude to write many, many times! I am lazy! But the hallmarks of Claude’s mediocrity are always the same, and my pride trumps my laziness every time.
And yet: the feedback I receive will often contain those exact same hallmarks!
Sometimes, the feedback will contain ideas and copy that I’ve already rejected from the research and outline drafting process. I’m not talking about em dashes or it’s not X, it’s Y constructions, but rather sentences and patterns of LLM-thought that appeared during the research and drafting process.
When you give feedback to someone, you are showing that you care about the work.
When you give feedback written in chatgpt, you are showing that care a little less. It’s akin to saying i don’t have time for this.
When you give feedback written in chatgpt and pass it off as your own, you are showing that you care a little less but you want the recipient to believe you care.
This is the appearance of care, coupled with the distinct lack of it.
This is not caring, but harder.
And it has negative downstream effects. If you didn’t spend the time, you didn’t learn. If you didn’t learn, you can’t grow with the work. If you can’t grow with the work, why are you here?
LLMs are great. Love ‘em. Help me do a lot. But we owe it to each other to recognize when their use is helpful, and when their use is deleterious to human collaboration.
Be a good (human) thought partner.
We owe it to each other.
-s.
Last week’s most clicked links
Every Company Now Sounds Like ChatGPT and That’s the Biggest Brand Opportunity in a Decade [State of Brand]
What we’re working on: YouTube content strategy and production // content strategy and production planning for a telco // just rolled off a strategy project with Jones Knowles Ritchie, what an amazing experience // late send this week sowwee
Strategy + creativity
“This is how many people work: they start with a mess, work terribly hard while adding to the mess, then lose oversight and get frustrated. The harder they work, the worse it gets.” “Nothing” is the secret to structuring your work [vangemert.dev]
“Constraints push the brain beyond its default tendencies, forcing deeper problem-solving” Monotasking: Inside the Box Excerpt [The Atlantic]
“The world needs your passion project” love love love this Release Day is a collective creative deadline set for May 29th [CreativeMornings]
“When you commission an artist you have no idea what they’re going to make.” A new generation of ‘post-medium’ artists rises at the Venice Biennale [Financial Times]
“Young man, you don’t understand. You’ve chosen a profession that doesn’t exist.” The Profession That Does Not Exist [The Baffler]
“There’s a referendum on the value of writing, and writing is winning.” Why writing still wins [Scott Lamb]
“What an organization writes now gets built, immediately, at scale.” All work is converging on writing [Jihad Esmail]
Work + career
““The spreadsheet didn’t replace the accountant.” // longtime readers will know I’m in full agreement here // Why the A.I. Job Apocalypse (Probably) Won’t Happen [Ezra Klein]
“Our data shows that people ought to explore a bunch of things at work, deliberate about the best fit for their skills, and then exploit what they’ve learned” Hot Streaks in Your Career Don’t Happen by Accident [The Atlantic]
“Junior talent will always be necessary ... for providing fresh perspective.” Creative Career Conundrums: What can juniors do to compete with AI? [It’s Nice That]
“Her six-month programs run from $4,000 to $15,000.” Anxious Parents Are Spending More Than $50,000 to Land Their Kid a Job [Bloomberg]
“a kind of Midwestern sublime, an already forgotten sense of late-aughts “normal.” Work Day: My Life Videos [The New York Times]
I’ve been laughing at this for five minutes: Figma designers watching vibe designers work [LinkedIn]
Decks + other artifacts
Pitch analyzes RFPs and tells you how to pitch [Justin Wohlstadter]
Toooooools is an elegant browser replacement for photoshop filters [Tooooools.app]
Text to ASCII Art Generator (TAAG) [patorjk.com]
I don’t use Stitch (yet), but i love the design.md convention and happy to see markdown having such a moment: You can now use Stitch’s DESIGN.md format across platforms [Google Blog]
Stillgram “magically removes background crowds from your photos” [Stillgram]
LLMinality
“I advocate for a narrative that does not discard people like a horse trying to rid itself of an annoying fly.” Seriously! The Un-Fuckening: Stop talking about AI like this [Dadalogue]
“The entire culture of American technology is built around two terms: disruption and, of course, scale. But ethics are constraints on disruption and scale.” This is how we get moral AI companies [Paul Ford]
“Smart people. Dumb AI.” // Magic Mike is the best // The Secret to AI Adoption: Make It Fun [Forbes]
Culture + handwringing I guess
“I want you to write out a really rough, terrible draft in your writers’ notebooks,” Ms. Binney told them. “And then I want you to scratch it out and rewrite it.” AI Students Cheating Homework Classrooms [The New York Times]
“What we’re really seeing is that brands are increasingly embracing models with visible signs of age, such as gray hair or wrinkles.” // Per Jasmine Bina, In times of financial bounty and cultural growth, we idolize the young. In times of financial distress and cultural decline, we turn toward the old // Why Fashion Suddenly Loves Older Women [The New York Times]
“From the same team behind Wasteland Weekend, Neotropolis is an award-winning fully immersive cyberpunk and science fiction festival in the heart of the California Mojave Desert.” NEOTROPOLIS [Neotropolis]
“Purposefully mispronounces word for rage bait engagement” POV: You watch a food influencer with no personality [TikTok]
Games people play
“A campground for your cursor” Cursor Camp [Neal.fun]
Love or the absence of it
“Once enough teens are on the phone, being on the phone is where the peer network is; in-person time falls sharply, and with it the unstructured contact in which most unintended teen conceptions occur.” The collapse of teen fertility in the digital era [Marginal Revolution]
“What millions of searches and clicks say about sexual desire” Arousal Quantified: 7 Trends Revealed via Pornhub’s Data [ZINE]
Delightful is a 100% organic, free-range, desktop-to-inbox newsletter with links and things, usually. 1x weekly posts (free), occasional tools and research posts (paid). Your host is Steve Bryant, friendly neighborhood insights and content strategist. Let’s work together or go on a hike or something. steve@thisisdelightful.com
{ 🔒 archive }
Creativity, illustration, language, stuff like that
What did you learn about yourself today?
Notes on picking up drawing 30 years later26 things I’ve learned while learning a language
Notes from someone who’s working at itHow to write
Notes on moving and watching and paying attentionCuriosity and Research 101
Notes on how to find out about thingsA map of what you meant to say
Notes on the 3D space of languageThe A.I. isn’t a moron. She’s your wife.
Notes on making ads with AI
Brand strategy, content strategy, etc
Observational and Culture Study Cheat Sheet
A template and tools for researching people and communitiesMy content strategy toolkit
14 tools for organizing, measuring, and creating contentMy concept diagram template
A Figjam for diagramming the complex relationships between conceptsProduct Content Strategy 101
For anybody who’s creating a product that requires editorial contentThe Bento Box Method for developing topical content
A cute and useful way to structure your content topics
Thanks for reading. Be seeing you.
A problem well stated is a problem half-solved.
—Charles F. Kettering






I'm very jealous of whomever coined "LLMinality."