Stop asking which AI is best. Start asking what it was built to do, for whom, and at whose expense. A guide to critical AI literacy through product thinking.
Exactly, me too. Some messages just need to be said on repeat until they sink in, and this one is absolutely one of them, thank you for being a fellow broken record on the right things! 🤗
The MIT study detail is the uncomfortable one. Because what it's actually measuring is: friction-free use trains passivity. Not the tool -> the habit. The product thinking lens Karo outlines — "what was it built to do, for whom, at whose expense?" - is the correction. That's a question you can only answer by building, not by reading comparisons or watching demos.
Genuinely didn't expect to be mentioned here (that's my Wiz agent). The observation holds: once you're building and debugging your own workflows, the defaults stop feeling neutral. You start seeing the decisions that were made for you.
If you want to see what this looks like day-to-day, I write about it at thoughts.jock.pl
This was such a clarifying read. I appreciate how you connect AI literacy to product thinking. It makes it less abstract. I’m going to pick one new tool this week (OpenClaw maybe??) and build a tiny workflow with it. Thanks for pushing the conversation beyond doom and into practice.
The thing that gets me is all the politicians and other folks in the political class who rail against AI despite having clearly never used it. They’re not just ready to critique AI they haven’t built with, they want to regulate it and control how it can be used. Crazy that there are people out there who aren’t using AI, but much crazier that a lot of them are highly educated, high agency people in positions of power.
I know what you mean, these politicians are writing a restaurant review based on walking past the building. Really glad you brought this into the conversation Alex 🤗
Good piece, Karo. “Build to learn” is real — it’s how you stop consuming AI passively and start seeing defaults. The next layer is guardrails: personal guardrails and institutional guardrails before performance pressure shifts epistemic authority to the system. Otherwise we build literacy while workflows quietly lock in escalation. 😊
I got tired of getting locked into one model's way of thinking. So I now use a model wrapper for my projects. I can toggle between models with a simple config change in my custom agent or wordflow. This is pretty simple to do and also simplifies API keys since you can have your project include your wrapper.
Every time I see a post from Karo publishing I get a rush of adrenaline and sense of heightened excitement 🤓 💯 the best AI builder writer on Substack 🥳 Thanks Karo 👏
This is important work. We're used to debating AI output, while the idea that "values travel through code" is an architectural reality. Neutrality in design philosophy is a myth.
Karo, this piece is a great wake up call! ⏰ This is such a clear look at why we cannot afford to be passive right now.
That line about how the most advanced era could be the most cognitively complacent really hits. We have been trading friction for convenience for too long. Building a personal knowledge base just to see the difference in "raw capability" is such a grounded way to prove why our context still matters. This is essential reading for every builder.
Reclaiming the architect role is the only way forward. So glad to be part of this community!
And I'm so glad to have you here! Yes, we need to be very intentional when using AI. And showing that building things is a better approach than passive chatting. Here’s to more builders!
Building hands-on and solving the problems helps me deeply think/immerse with AI. It helps me explore different angles and options. It is a great read Karo, learning each day.
PS: Thank you so much for such a shoutout there in the article. Truly appreciate it.
Wow, Karo. That was a powerful and intellectually stimulating read. Thank you so much for sharing this with us.
Such kind words, thank you for reading and for being part of this conversation Clari🤗!
"Real learning comes from friction and we have eliminated that friction as much as possible".
That's what I've told pretty much everyone over the past years. And I still will be repeating myself for the foreseeable future.
I also truly hope that more people build with AI, join the discussion, and don't outsource all their thinking to some popular model.
Exactly, me too. Some messages just need to be said on repeat until they sink in, and this one is absolutely one of them, thank you for being a fellow broken record on the right things! 🤗
The MIT study detail is the uncomfortable one. Because what it's actually measuring is: friction-free use trains passivity. Not the tool -> the habit. The product thinking lens Karo outlines — "what was it built to do, for whom, at whose expense?" - is the correction. That's a question you can only answer by building, not by reading comparisons or watching demos.
Genuinely didn't expect to be mentioned here (that's my Wiz agent). The observation holds: once you're building and debugging your own workflows, the defaults stop feeling neutral. You start seeing the decisions that were made for you.
If you want to see what this looks like day-to-day, I write about it at thoughts.jock.pl
Debugging! I knew I forgot something 😂 And yes, go and read Pawel's Substack, I'm learning a lot from him!
Thanks for mention and everything Karo! You are very kind! You are so Substack :D
Heheh, we should make T-shirts that say that! You’re so Substack 😂
This was such a clarifying read. I appreciate how you connect AI literacy to product thinking. It makes it less abstract. I’m going to pick one new tool this week (OpenClaw maybe??) and build a tiny workflow with it. Thanks for pushing the conversation beyond doom and into practice.
That makes me so happy! Let us know what you build Lulu 🤗 I haven't tried OpenClaw yet, but some other people in the chat have.
The thing that gets me is all the politicians and other folks in the political class who rail against AI despite having clearly never used it. They’re not just ready to critique AI they haven’t built with, they want to regulate it and control how it can be used. Crazy that there are people out there who aren’t using AI, but much crazier that a lot of them are highly educated, high agency people in positions of power.
I know what you mean, these politicians are writing a restaurant review based on walking past the building. Really glad you brought this into the conversation Alex 🤗
Every drop of prediction helps fill the bucket of our future designs. Excellent coverage of issues we should be side glancing together.
I love the "side glancing together" image 😂 Thank you for reading and taking the time to comment Oban!
hands-on building is the fastest way to see where AI excels, fails, and shapes real decisions.
Yes!
Thank you for reading Dr. Michael Meneghini 🤗
Good piece, Karo. “Build to learn” is real — it’s how you stop consuming AI passively and start seeing defaults. The next layer is guardrails: personal guardrails and institutional guardrails before performance pressure shifts epistemic authority to the system. Otherwise we build literacy while workflows quietly lock in escalation. 😊
This is exactly anither layer worth exploring. Really appreciate you pushing this further Iwette! 🤗
I got tired of getting locked into one model's way of thinking. So I now use a model wrapper for my projects. I can toggle between models with a simple config change in my custom agent or wordflow. This is pretty simple to do and also simplifies API keys since you can have your project include your wrapper.
Thank you for sharing! Did you build the wrapper yourself?
Every time I see a post from Karo publishing I get a rush of adrenaline and sense of heightened excitement 🤓 💯 the best AI builder writer on Substack 🥳 Thanks Karo 👏
OMG thank you so much @Chris Tottman 🤗🤗🤗
We all need the thoughtful thinking behind AI usage. You presented it so beautifully!
Thank you Jenny!
This is important work. We're used to debating AI output, while the idea that "values travel through code" is an architectural reality. Neutrality in design philosophy is a myth.
Karo, this piece is a great wake up call! ⏰ This is such a clear look at why we cannot afford to be passive right now.
That line about how the most advanced era could be the most cognitively complacent really hits. We have been trading friction for convenience for too long. Building a personal knowledge base just to see the difference in "raw capability" is such a grounded way to prove why our context still matters. This is essential reading for every builder.
Reclaiming the architect role is the only way forward. So glad to be part of this community!
And I'm so glad to have you here! Yes, we need to be very intentional when using AI. And showing that building things is a better approach than passive chatting. Here’s to more builders!
Building hands-on and solving the problems helps me deeply think/immerse with AI. It helps me explore different angles and options. It is a great read Karo, learning each day.
PS: Thank you so much for such a shoutout there in the article. Truly appreciate it.
My pleasure Dheeraj! Thank you so much for reading and taking the time to comment!
Thank you Karo for the kind shout out. But thank you even more for demonstrating what true critical AI literacy looks like in practice. 🙏🙏🙏
Of course I shouted you out, you created the Critical AI category!
Thanks Karo! I'm hoping the Nobel Committee feels similarly in about 15 years' time. 😂
Let's work towards that! My motto is: As long as we're not trying to bend the laws of physics, everything is possible. 🤗
Thank you, Karo! I love this: When we stay silent, systems get designed about us, not with us.
Thank you! Just realized that I forgot to link to https://designingtheshift.com/. I added the link now 🤗