This framework brilliantly reframes AI from a shortcut into a thinking partner, showing that deliberate engagement can actually strengthen critical thinking rather than replace it.
Yes! Many of us are experimenting with different ways to extend, rather than replace, ourselves. This is an area that doesn't get as much attention as it should, but it's exciting to see. I think it is going to be one of the real advantages of AI in the future.
I recently built out a framework myself that seeks to not only elicit the best thinking you have to offer, but capture and compound it. It's called the Cognitive Mesh Architecture:
The Cognitive Mesh is based on a multi-agent architecture on a new platform (brightsy.ai), but the goal is that it's not platform-specific. Here's how we implemented it for the StoryCycle Genie™ (detailed in the whitepaper and outlined in the article above). Hopefully, this will provide some clarity. It's new and evolving, but it exists and is working well so far.
1. Started with a framework to create a process that is both generative in nature and designed for human-in-the-loop interaction to create high-fidelity outputs.
2. We leverage UX and behavioral design principles to create the conditions for cognitive flow. For example, dynamically inserting contextually relevant UI elements such as buttons, forms, charts, etc. If we're successful, the human-in-the-loop will be unburdened by cognitive load, and the thinking comes easily.
3. Once the output is validated as high quality by the human (due to being a part of the process), the output is saved in the system with semantically intelligent metadata.
4. Now, when the next specialized agent is used, it will have access to this ultra-intelligent content (more to come on that) and can use it to inform the next generative process.
5. The process repeats, compounding the intelligence of multiple people, adding to the corpus of outputs, each with their own intelligence incorporated into the outputs, creating a collective intelligence ecosystem.
I'll be exploring how everything works in the coming weeks, including how it can be adapted for various industries.
Excellent thoughts, Daria. In these kinds of "big picture" discussions, putting one's positionality with AI at the forefront is a prerequisite to practicing specific techniques. It's a bit like the old "teach someone to fish" proverb: If we give students too many "fish" before they learn to fish for themselves, it will unintentionally imprint the idea that students are clients of AI rather than operators of it. Your prompts reinforce the positionality as operator.
Daria, I was wondering if the main risk isn't AI making us dumber, but allowing it to make us more certain. Without these deliberate pauses, we get polished, confident, and wrong answers faster than ever. Your framework serves as a helpful defense mechanism. It's a supportive way to harness technology for a clearer understanding of questions.
Such a great point, I hadn’t think about it as “false certainty” before, but you’re so right, that’s a real risk with AI. Really appreciate you adding that perspective and thank you for your words!
Since I spend 98% of my online time on devises, it may be something to consider. Even a pdf of the prompts for subscribers. That could be very beneficial! I’ll try it on my iPad too. When there’s a will, there’s a way!!
Love this! The “think first, ask better, challenge the answer” approach really hits home. AI feels less like a replacement and more like a thinking partner when we engage intentionally. Excited to try Socratic Questioning, Devil’s Advocate, and Reframing Setbacks in my work.
Love this! I got into some serious trouble due to AI brain during my last holiday, so great to read about ways to fight back :) https://metacircuits.substack.com/p/is-your-brain-over-reliant-on-ai
“AI brain” is a great way to put it:)) I’ve felt that too. I let AI plan a holiday once, and it completely messed up the experience for me 😅
Avoiding AI isn’t principled—it’s opting out. The real work is building practices that keep us intellectually awake while we use them.
What’s one practice you won’t compromise on?
The framework I outlined in the article. What's yours?
Its not possible to replace your mind unless you are able to perform transmigration.
Not in the literal sense haha, more like outsourcing thinking
Great post! Love the framework. Using AI to critique and questions things is my favourite use case at the moment. Thanks for sharing your prompts. ❤️
Thanks so much, Katrina! Critiquing + questioning is my most used case too
Love the list Daria! Socratic is my fav!
Thanks Wyndo, Socratic is such a good one! Timeless...
Love the 17 prompts! Simple but effective.
Thank you, Jenny!
Featured this in today's roundup :) https://allisonstadd.substack.com/p/sniffaris-t-shirt-wisdom-and-the
thanks for the shoutout, Allison!
This framework brilliantly reframes AI from a shortcut into a thinking partner, showing that deliberate engagement can actually strengthen critical thinking rather than replace it.
Thank you, Suhrab!
Yes! Many of us are experimenting with different ways to extend, rather than replace, ourselves. This is an area that doesn't get as much attention as it should, but it's exciting to see. I think it is going to be one of the real advantages of AI in the future.
I recently built out a framework myself that seeks to not only elicit the best thinking you have to offer, but capture and compound it. It's called the Cognitive Mesh Architecture:
https://thecognitivemesh.substack.com/p/the-cognitive-extension-opportunity?r=g1nmt
This is a very interesting development. How could someone actually implement this in practice?
The Cognitive Mesh is based on a multi-agent architecture on a new platform (brightsy.ai), but the goal is that it's not platform-specific. Here's how we implemented it for the StoryCycle Genie™ (detailed in the whitepaper and outlined in the article above). Hopefully, this will provide some clarity. It's new and evolving, but it exists and is working well so far.
1. Started with a framework to create a process that is both generative in nature and designed for human-in-the-loop interaction to create high-fidelity outputs.
2. We leverage UX and behavioral design principles to create the conditions for cognitive flow. For example, dynamically inserting contextually relevant UI elements such as buttons, forms, charts, etc. If we're successful, the human-in-the-loop will be unburdened by cognitive load, and the thinking comes easily.
3. Once the output is validated as high quality by the human (due to being a part of the process), the output is saved in the system with semantically intelligent metadata.
4. Now, when the next specialized agent is used, it will have access to this ultra-intelligent content (more to come on that) and can use it to inform the next generative process.
5. The process repeats, compounding the intelligence of multiple people, adding to the corpus of outputs, each with their own intelligence incorporated into the outputs, creating a collective intelligence ecosystem.
I'll be exploring how everything works in the coming weeks, including how it can be adapted for various industries.
Is that helpful?
Much more clear now! Very interesting project.
Thanks!
I never ask AI anything that I haven't brainstormed by myself first. It allows me to find mistakes better.
Amazing approach, taking time to clear your mind first is exactly the kind of slowdown we need more of!
Excellent thoughts, Daria. In these kinds of "big picture" discussions, putting one's positionality with AI at the forefront is a prerequisite to practicing specific techniques. It's a bit like the old "teach someone to fish" proverb: If we give students too many "fish" before they learn to fish for themselves, it will unintentionally imprint the idea that students are clients of AI rather than operators of it. Your prompts reinforce the positionality as operator.
Here is a "Spectrum of AI Engagement" that I think helps to codify where we are all at, as operators of AI: https://stevecovello.substack.com/p/why-do-we-assess-reflection-in-an
Thanks Steve, really like that point. Being an operator, not a client, makes a lot of sense. I’ll add it to my weekend reads, looking fwd to it.
Daria, I was wondering if the main risk isn't AI making us dumber, but allowing it to make us more certain. Without these deliberate pauses, we get polished, confident, and wrong answers faster than ever. Your framework serves as a helpful defense mechanism. It's a supportive way to harness technology for a clearer understanding of questions.
Such a great point, I hadn’t think about it as “false certainty” before, but you’re so right, that’s a real risk with AI. Really appreciate you adding that perspective and thank you for your words!
Hi Daria! Great article and prompts. Is it because I’m on my phone that I can’t copy the prompts? Thank you!!
Oh yes, that’s probably it, the prompts are in a code block and it seems they can only be copied from a computer. Sorry about that, I had no idea
I was able to open Substack in my browser on my iPad and copy the descriptions and prompts completely in one copy and paste!
Awesome!
Since I spend 98% of my online time on devises, it may be something to consider. Even a pdf of the prompts for subscribers. That could be very beneficial! I’ll try it on my iPad too. When there’s a will, there’s a way!!
Will surely do! If it doesn't work from you iPad, let me know, and I'll send you a PDF.
I have some ideas. The iPad is next on my to do list. ✅
Love this! The “think first, ask better, challenge the answer” approach really hits home. AI feels less like a replacement and more like a thinking partner when we engage intentionally. Excited to try Socratic Questioning, Devil’s Advocate, and Reframing Setbacks in my work.
Intentional engagement makes all the difference. Excited to hear how those prompts work for you 🤗
Loved your prompt
Thank you, Farida! Curious what are your favorites / most-used ones
Stress test and Socratic will be my favorite
Great picks :)
Nice! Saved the prompts for later ;)
Let me know how they work for you!