Damn, I read it twice already. And saved it 😂 So many wisdom moments in one article. Your newsletter is probably the one that makes me think the hardest. Thank you for that!
What I really like about this is that you’re describing almost the opposite of cognitive offloading.
You’re not asking AI to do the thinking for you. You’re forcing it to expose why it reached a conclusion, what assumptions came with the category, and where that category stops explaining the actual material.
That creates more work for the human judgment part, not less.
The line “my job is to decide what belongs” really landed for me. The value isn’t that AI gives you an answer. It’s that you can keep challenging the answer until your own thinking gets sharper.
Thank you for reading Gabe! :) I also think there's a suuuper fine line between having AI to think for you and yourself doing the thinking. It's super tricky, and we don't really have a manual or definite set of instructions for it.. but what we can do is try our best to not fall for the temptation. At least not all the time! 😁
Very insightful article! I will stick to my old fashioned routine of drafting my article on the computer without AI, revising it a few times, and then putting it through the spell checker. I use AI off and on, and end up wasting precious time trying to correct it!
Thank you Hema, this is the perfect plan! I also don't let AI draft or write - I use it to edit and even then I never copy its suggestions line by line. Correcting it does indeed take up SO MUCH precious, creative time ❤️
I really like the two example prompts. I find myself fighting the early choice of category fairly often. Like, whoa bruh, you don't know me that well yet.
And honestly, I think fighting it early is super good, as once AI gets too confident too quickly, you’ll then spend the entire convo undoing the personality it invented for the work.
So timely. I am working on an interview series and haven’t been able to tame my LLM to get the level of output I want. So I have been writing my series myself end to end. This gave me food for thought and I am going to try with a changed approach now.
Mia, finding this article made me literally squeal with joy because it explores so well something I've spent a lot of time wondering about, and I even posted about it here https://butwhyai.substack.com/p/how-dare-you-delete-it.
You do expand more on one of the angles I touch on (naming something prematurely) and I go into other directions, but at the core, it's AI categorising things fast and loose.
Such an enjoyable read, Mia! I came to using AI relatively recently, and I've been thinking a great deal about how to make the most of the collaboration without handing over too much of the thinking I want to keep practicing. I love your distinction between what a category can help us see and what it may prematurely close off. There is something so important in keeping the human work of deciding what belongs - and remaining curious about what doesn't. You've given me a lot to think about here. I look forward to reading more of your writing!
Meg, hiiii! And welcome to the robots world ❤️ I root for the human. I'm not sure I know where this is all going in 5 years, but I want to make sure we're prepared for the "now".
I love the idea of giving the strange bits more time. That might be one of the best instructions we can give both AI and ourselves. I hope you are having a good week sis :)
I tried to brainstorm a name for my publication with Astra. I’m not happy that everyone seems to have such great names, while mine is just my full name. It turns out that the best names are already taken, and I had high hopes for Astra’s help.
It has a lot of context: all my essays, my Notes, even plans for future pieces. Still, none of the names clicked. Its suggestions gave me a noun based on what I had already published. But the name I was looking for needed to come from a still-unformed idea of what I plan to do next.
This failed brainstorm did help me understand why nothing felt right. I haven’t decided yet what my content strategy will look like or what the main theme will be.
So yes, the decision stays with us.
And AI can make the gap between what’s settled and what’s still unformed very clear.
In my experience, AI is pretty bad at coming up with names! You do need hundreds of iterations, and context, and at that point, it's better to sit down in silence for a few hours and it'll come to you ❤️
Lovely read and, as always, so refreshing. Thank you
As I was listening to it (and by the way I usually listen to the articles), Substack's voice drives me nuts so I'm using Speechify. Sorry, side note
I was thinking how compartmentalization can sometimes come in handy. For example when you tell it to imagine it's a world-class UX researcher
we also use this mechanism sometimes in brainstorming: one may think that having those anchors may limit people's ideas and one might think it constrains or limits them but sometimes it provokes creativity
Super interesting, thank you so much for your perspective! It makes a lot of sense.
It used to work with robots too, for instance it used to be best practice to tell AI "act as a worldclass copywriter and..." It would do the job pretty well!
I think stuff changed about a few months ago, and it looks like you're better off not saying that anymore, which I find super interesting!
On your side note - I appreciate you telling me this because I'm considering narrating the posts myself soon! ❤️
Damn, I read it twice already. And saved it 😂 So many wisdom moments in one article. Your newsletter is probably the one that makes me think the hardest. Thank you for that!
BEST COMPLIMENT EVER ❤️
What I really like about this is that you’re describing almost the opposite of cognitive offloading.
You’re not asking AI to do the thinking for you. You’re forcing it to expose why it reached a conclusion, what assumptions came with the category, and where that category stops explaining the actual material.
That creates more work for the human judgment part, not less.
The line “my job is to decide what belongs” really landed for me. The value isn’t that AI gives you an answer. It’s that you can keep challenging the answer until your own thinking gets sharper.
Thank you for reading Gabe! :) I also think there's a suuuper fine line between having AI to think for you and yourself doing the thinking. It's super tricky, and we don't really have a manual or definite set of instructions for it.. but what we can do is try our best to not fall for the temptation. At least not all the time! 😁
Very insightful article! I will stick to my old fashioned routine of drafting my article on the computer without AI, revising it a few times, and then putting it through the spell checker. I use AI off and on, and end up wasting precious time trying to correct it!
Thank you Hema, this is the perfect plan! I also don't let AI draft or write - I use it to edit and even then I never copy its suggestions line by line. Correcting it does indeed take up SO MUCH precious, creative time ❤️
I really like the two example prompts. I find myself fighting the early choice of category fairly often. Like, whoa bruh, you don't know me that well yet.
Yesss exactly the energy! 🤣
And honestly, I think fighting it early is super good, as once AI gets too confident too quickly, you’ll then spend the entire convo undoing the personality it invented for the work.
Agree. 100%!
So timely. I am working on an interview series and haven’t been able to tame my LLM to get the level of output I want. So I have been writing my series myself end to end. This gave me food for thought and I am going to try with a changed approach now.
I’m all for that!! ❤️❤️❤️
Mia, finding this article made me literally squeal with joy because it explores so well something I've spent a lot of time wondering about, and I even posted about it here https://butwhyai.substack.com/p/how-dare-you-delete-it.
You do expand more on one of the angles I touch on (naming something prematurely) and I go into other directions, but at the core, it's AI categorising things fast and loose.
Fantastic, I loved this :) you've got a new sub ❤️
🤩🤩🤩 so exciting - thank you! What an honour!
The honour is mine ❤️🥰
Such an enjoyable read, Mia! I came to using AI relatively recently, and I've been thinking a great deal about how to make the most of the collaboration without handing over too much of the thinking I want to keep practicing. I love your distinction between what a category can help us see and what it may prematurely close off. There is something so important in keeping the human work of deciding what belongs - and remaining curious about what doesn't. You've given me a lot to think about here. I look forward to reading more of your writing!
Meg, hiiii! And welcome to the robots world ❤️ I root for the human. I'm not sure I know where this is all going in 5 years, but I want to make sure we're prepared for the "now".
I love the idea of giving the strange bits more time. That might be one of the best instructions we can give both AI and ourselves. I hope you are having a good week sis :)
So true Neela! And I am, thank you ❤️ all by the waves haha 🌊 hope you are too!
I am better that today is Friday :)
Have a good weekend sis
I create rules to prevent AI from prematurely rejecting or deciding. It takes a lot of cycles to find those rules.
As you should!
It’s a mentorship I didn’t want or ask for but at least it’s entertaining
This is a story about my evening yesterday. :)
I tried to brainstorm a name for my publication with Astra. I’m not happy that everyone seems to have such great names, while mine is just my full name. It turns out that the best names are already taken, and I had high hopes for Astra’s help.
It has a lot of context: all my essays, my Notes, even plans for future pieces. Still, none of the names clicked. Its suggestions gave me a noun based on what I had already published. But the name I was looking for needed to come from a still-unformed idea of what I plan to do next.
This failed brainstorm did help me understand why nothing felt right. I haven’t decided yet what my content strategy will look like or what the main theme will be.
So yes, the decision stays with us.
And AI can make the gap between what’s settled and what’s still unformed very clear.
In my experience, AI is pretty bad at coming up with names! You do need hundreds of iterations, and context, and at that point, it's better to sit down in silence for a few hours and it'll come to you ❤️
This was really insightful. Thank you.
Thank you for reading! 🙏
Lovely read and, as always, so refreshing. Thank you
As I was listening to it (and by the way I usually listen to the articles), Substack's voice drives me nuts so I'm using Speechify. Sorry, side note
I was thinking how compartmentalization can sometimes come in handy. For example when you tell it to imagine it's a world-class UX researcher
we also use this mechanism sometimes in brainstorming: one may think that having those anchors may limit people's ideas and one might think it constrains or limits them but sometimes it provokes creativity
I guess it’s a different story for robots
Super interesting, thank you so much for your perspective! It makes a lot of sense.
It used to work with robots too, for instance it used to be best practice to tell AI "act as a worldclass copywriter and..." It would do the job pretty well!
I think stuff changed about a few months ago, and it looks like you're better off not saying that anymore, which I find super interesting!
On your side note - I appreciate you telling me this because I'm considering narrating the posts myself soon! ❤️
Oh I'm clearly out of the loop. Do you know why we don't use the world-class copywriter anymore? 😅
World love your narrated articles ❤️
As far as I know, it's because AI got smarter and you don't need to give it a "role". But there might be smarter reasons too hahaha.