Scribbling and thinking // Writing against AI
First, a quick riff on creativity. Then some thoughts on writing and content in the age of ChatGPT.
Ralph Ammer’s notes on how to draw ideas align nicely with my own approach to writing/thinking: lots of messy, throwaway scribbles, doodles, and improvisations. May your waste paper basket overfloweth:
Now, let’s talk AI.
ChatGPT is really, really good. Try it and be amazed. It’s one of the reasons I’ve been writing less. I don’t want to write anything that could have been written by the AI, and it can now write a lot of things very well. Any summary of existing knowledge is absolutely worthless as human-created content: ChatGPT can do it better.
So what still works? And what will work into the future?
I think Ralph Ammer’s approach, above, is a good one: longer form, more detailed, and multi-media. Raise the production value. He uses animations, doodles, diagrams. His stuff is great.
Another approach, related to the topic of creativity, is to have fresh ideas. Paul Graham is probably the paragon of this, with lots of interesting, new ideas. The other day, I spent six hours working through a single one of his essays (this one), because it was prompting so much thinking in me.
ChatGPT is incredible at answering any sort of “best practice” question (or to generate cookie-cutter “ideas” like this), but I wasn’t able to get it to be truly novel. It’s certainly going to replace a lot of “answer to this question” SEO articles, where a smart language model, similar to ChatGPT, will simply replace 90%+ of search results without ever sending you anywhere. (And that sort of content was pretty meaningless and humanity-destroying to “write,” anyway, so no great loss.) So writing returns, hopefully, to being about real new ideas.
A third approach might be to go the opposite direction with style and aesthetics. As AI’s prose approaches formal perfection, we can go avant garde. What’s the Dada or Cubist take on content marketing? Might be time for a re-read of The Work of Art in the Age of Mechanical Reproduction (1935).
Or maybe wrap more of your personality into it. More video. More vulnerability. More humanity. More of you. I don’t necessarily think that any one of these is enough, on its own, since any can be fairly accurately modelled. But maybe in combination, being less polished and more you across the board, might be viable. (Also: humor.)
Lastly, and I think maybe the most practical for most situations, is to niche down real hard. Language models are, by definition, a bit generic. They give a good answer for the average person, but most people aren’t average. If there’s a way to speak just to them, directly enough that it stands head and shoulders above other advice for that paticular type of person, then maybe that works.
Comments (12)
in the past, i would always force through the study and explore stages.
As if they need to start and end on time so that the develop and show can happen on schedule.
I've now realized and accepted that the study and explore need to take as much time as they need to and there's very little I (or at least the conscious part of me) can do about it.
Funnily enough
Here’s a tweet by John Cutler for the context of a PM in an organization where there’s low stability/certainty
But I feel this applies to solo person as well
Instead of just heads down go and build more and more complex plans on the outside without any strong foundations on the inside, I prefer to build foundations on the inside via exploration mode
Even if that means I have no certainty what exploration may yield or how long it takes
https://twitter.com/johncutlefish/status/1599141256541896704?s=20&t=lIVnbrevhWICRcfW43UGYw
[IMG_8251.jpeg]
Rob, how nice of you to lay the bar to the noblest height: refuse to write what AI can come up with. And thanks for the hopeful alternatives! If you see yourself 'exploring' five years from now, which one would dominate?
For myself, anything that embraces the physical human body to explore ideas will probably the most lively option.
Let your body experience the idea! Let your pen bleed, write without stopping for ... minutes! Thanks Ralph Ammer! I like artist Austin Kleon's view on this in this lovely podcast. Some call it mindstorming, and then let your brain process it while you sleep, see what's left of the exploration after your mind's done with it. Mimic the physical act that your idea requires.
I think the musician Paul Weller might have done some brainstorming on the supposed question he asked himself ,,Say Paul, what is entertainment, actually?" He must have written a vast vast list (you know, if you see people like Peter Blake around you, collaging everyday life and making it art), then went to sleep, raked up the ideas that his deep sleep had sieved and shaped for him and crafted them neatly into this song. Dadaists might turn their idea exploration in a song or a painting.
As a recovering academic (any others out there?) I try to actively experience ideas and idea exploration away from digital formats. So, note taking with a pen and paper, drawing, building something out of lego or match stick boxes. It feels more uncomfortable, ha!, but definitely more fun and in the end 110% wiser. Everyday that goes by here in the digital age, where we don't question typing, watching images, video's could be a sorry waste. Niche yourself down, why don't you, to a physical idea explorer.
If I had to make a near-future prediction, I think there's going to be a super annoying, zero-sum-game-at-scale where a lot of people are "writing" a lot of content, without realizing that they're doing so in pursuit of winning an SEO/audience game that's already in its final days. E.g., people are already trying to "win" at stack overflow like this, with predictable results. They're all playing yesterday's game, IMO. Better to learn how to have interesting ideas. Or at least ask interesting questions. Or to get back to the craft and art and humanity of writing, instead of its current clinical format...
PG has an article on coming up with startup ideas where the interesting meta-conclusion is that if you rush the process, you end up with an idea that seems good, but isn't. So a lot of his advice is about putting yourself in a place and mindset where you're able to notice, and then just go on living. It's hard advice to take, cuz it's like, "But I need an idea NOW!" But seems like correct advice...
Wow! I had no idea how missused this ChatGPT actually is. I guess its missuse comes when people view it as a “main lever” instead of a helpfull tool. What a weird sign of the times. ,,In France a skinny pen died of a big disease with a little name (ChatGTP)”. Sorry, Prince!
Seeing the reply-thread there is hopeful. But we’re warned now, right?
Haha,
,,You can go your own way. Go on your own way”, more songs! We always could and we always had to.
And if the SEO game is in its final days, then rightfully, again, actual human creativity will have its day once again. It’s good to see that this here is a safe haven. Here, things always move forward, happy to tag along. Thanks!
I recently bought a book that helps you with a 110% AI-free activity, meeting people, so nice, I’m diving in and doing it!
Peace, out!
interesting
I have read that article before but I have never read it the way you did
This is thought provoking
I wonder by reading it the way you did and taking it seriously what behaviors will I change going forward?
One for me has been treating side projects as exploration of idea spaces (and cofounder collaborations). Important seeds to plant, in both cases, for the long game.
Hi , new here... and I see we are closing up shop (on the subscription side) but that is okay! I totally get it.... Sharing something that you might appreciate. GPT is totally useless at writing anything actually readable and meaningful (unless iterating with a writer in which case it is an okay assistant)... https://mailchi.mp/943e11baae5b/introducing-in-the-loop-9322250?e=b3ec66545c
> don't want to write anything that could have been written by the AI, and it can now write a lot of things very well.
Yes this is for me a great motivating factor as well
Strangely I’m more excited now that the bar is raised on me.
Not that I was writing very well or often enough in the first place
When I read this
> third approach might be to go the opposite direction with style and aesthetics
I think of Nassim Taleb’s uh annoying writing style 😅
I recently wrote these 3 things as more important given chatGPT raising the bar
I guess adding style and aesthetics should be number 4
‘’’
My guess is the following are going to be even more important than ever before:
1️⃣ Quality of the output (writing/video),
2️⃣ hard-to-fake signals like receipts,
3️⃣ tight fit between creator-topic-audience
banal generalities hollered out in any direction are going to fail
‘’’
Original source:
https://twitter.com/KimStacks/status/1638178696912846859
this is so good rob.
• sketches - ralphammer +1
• video w/humor
I've been dissecting pg's new article on how to do great work and it's fun to triangulate with the work you live piece.
https://www.notion.so/mcandler/How-to-Do-Great-Work-a763041fa75a42c583e5aabc07352573?pvs=4