Is AI Making Me Think Less? [Raw Session]

00:00 — When asking AI becomes the first step

Hey, welcome back to Slow Builds.

I've been thinking about something with AI that feels a little uncomfortable, mostly because it involves how I'm using it myself.

I use AI to help me with writing, write these scripts. It helps me format it, segment it, do the websites. It helps with coding, helps with the YouTube channel, like when it comes time to thumbnails, titles, descriptions, comments, replies. And sometimes just figuring out what I'm trying to say, and for text messages, emails. And I get a lot of value from that.

But I've started wondering what happens when asking AI becomes the first step.

We did go from an era of when we had to look everything up. My friend and I were talking a while ago about how we used to have to go to the encyclopedia to look something up. But then that turned to AltaVista or—Lycos, all the old websites, and the original Yahoo. And then Google changed that.

But now AI has become the first step in almost everything. Before I've tried to explain the problem, before I've decided what I think, before I've even sat with the question for a few minutes in my head, trying to figure out what I think, what I think the answer is, or what the outcome is, or trying to remember something from the past.

I'm not saying everyone is going to forget how to think. I'm wondering about something quieter: whether I could get so used to having help with the first pass, the first instinct, that I stop wanting to do it myself.

01:49 — What would AI limits actually measure?

The easiest comparison for me is phones. They became useful enough and easy enough to keep reaching for that we started putting limits around how we use them. Screen time, app time, app limits, focus mode, sleep mode, Do Not Disturb. All these things are ways of putting a little friction back into something we've made incredibly convenient.

I wonder what the equivalent looks like for AI. Are we going to see that in AI? Is AI going to tell us to take a break? Is AI going to say, "Maybe you should go read this, go find this, go here, step away"?

Maybe it's going to do this. Maybe we can put limits in it. Maybe we can, just like the screen time, or kids, or anything else.

Because it may not be about how many minutes I spend using it. I could spend an hour with AI questioning an idea and thinking really hard, or I could spend thirty seconds accepting an answer I barely understand.

The time alone wouldn't tell me much. I'd need to look at what I was actually doing during that interaction.

03:02 — This video started with AI too

And I need to be clear: I'm not standing outside of this. I use AI for these videos. I use an idea. We build out on that idea. We create a script for it. And then after the script is done, then we see if there's a title, a description, first comment, and then a thumbnail idea.

Then we move on to rewriting the script, because I'll read it, we tweak it, we change it. And then once the videos are up, AI helps me convert that for the website. It helps me do the thumbnail. Now I'm using it to help me do Shorts, because that's a new thing that I've discovered, and I kind of like them. Hopefully a few come out of this one.

But anyway, sometimes I bring a thought that's barely formed, and I ask it, "Help me work out what's there." So yeah, this video started that way.

The rough thought was something like, "AI causes people to lose reasoning." And even saying that now, it sounds more certain than what I actually mean.

I don't know that I'm losing my ability to reason. What I'm trying to understand is whether the convenience could change my habits. Those are different claims, and working through that difference is part of the thinking.

When I read a draft and realize it's saying something I don't actually believe, I have to stop and correct it. Sometimes I reject a whole direction. Sometimes the wording sounds good, but it makes the idea too certain or misses the part I was actually interested in. And that's still reasoning.

So I don't think it would be honest to say that using AI means I've stopped thinking. The question is whether I'm staying involved, and whether I'd notice if that started to change.

05:06 — A polished answer can feel finished

One thing that makes this difficult is how quickly AI can give a thought a finished shape. A rough idea goes in, a clear explanation comes back with a beginning, a middle, and a conclusion. And that can be useful.

Sometimes seeing the idea written out is exactly what helps me understand what's wrong with it. But I think there's also a temptation to accept the shape because it feels complete. The sentences connect. The explanation sounds reasonable. Nothing immediately jumps out as wrong.

And suddenly I'm reviewing the answer it gave me instead of working out how I would have approached the question myself. And that doesn't automatically make the answer worse, but it can influence where my thinking starts and which possibilities I consider.

If it gives me three options, do I seriously consider that there might be a fourth? If it frames something as a problem, do I stop to ask whether I agree with that framing? Or do I just start improving what's already in front of me?

That's the part I want to pay attention to. AI can make weak thinking look finished. And I don't want to confuse a clean explanation with having understood something.

06:23 — What do I want to stay good at?

I also think there are different kinds of work happening here, and I don't want to lump all of them together.

If I know what I want to say and AI helps me shorten a sentence, that's one thing. If I'm trying to make a decision and I ask what I should value, that's something else.

I don't need to personally struggle with every title, description, or every piece of routine writing to prove I'm still capable. Some of that work is exactly what I want help with. The harder question is which parts I actually want to remain good at myself.

If I care about explaining my ideas, then I probably need some practice explaining them. If I care about judging an argument, I need to do more than recognize that it sounds convincing. And if I'm making a choice that affects my life, I want to understand why I'm making it.

It doesn't mean doing everything alone. It means being a little more deliberate about what I'm handing over.

07:21 — Working through uncertainty or making it disappear

Because the early part of thinking can be uncomfortable. You don't quite know what you mean. You try to explain it, and halfway through you realize the pieces don't fit. You write something down, read it back, and discover you've made a claim you can't really defend.

I don't enjoy all of that. But sometimes that's where I find out what the actual question is.

AI can help with that process. It can ask something I hadn't considered or show me a contradiction I wasn't seeing. And it can also give me a smooth way around the discomfort.

And I'm not sure I always know which one I'm asking it to do. Am I asking for help working through the uncertainty, or am I asking for something that makes the uncertainty feel finished?

Those could look very similar from the outside. In both cases, I end up with a readable answer. But I might come away understanding very different amounts.

08:26 — Losing the urge to make my own first attempt

The part that worries me most is the possibility that I stop wanting to make my own first attempt. Because my first attempt is probably going to look worse. I'll be slower. It'll have holes. I'll repeat myself.

AI can give me something that looks much further along almost immediately. Once that's available, starting from scratch can feel unnecessary. And sometimes it is. It is unnecessary. But I don't want speed to make those decisions for me every single time.

There are thoughts I like to sit with before somebody or something starts organizing them for me. I might take a different direction. I might realize I'm asking the wrong question. Or I might end up in roughly the same place, but understand how I got there.

I think there's value in that, even when the finished result doesn't look more impressive.

09:17 — Can I explain why I agree?

So for me, a useful version of AI isn't necessarily the one that does the least. It's the one that helps me understand more.

Sometimes that means bringing a rough thought first. Even a few sentences about what I currently think, what I'm unsure about, and what I'm trying to work out. Then I've got something to compare its response against.

Did it actually address my concern? Did it change my mind? Or did it just say something more confidently than I did?

Other times I might need help getting started. I don't think that should somehow disqualify the thinking that happens afterwards. A draft can give me something useful to push against. But I want to go beyond whether I like the wording.

Can I explain the main idea without reading it back? Can I say why I agree? Can I point to something that would make me reconsider?

Those questions seem more useful than counting how many words I wrote myself. Because changing a few sentences doesn't necessarily mean I've explained the argument. And accepting a suggestion doesn't necessarily mean I haven't.

What matters is what I understand and decide along the way.

10:34 — The kind of help I want from AI

That's also the kind of support I want from the tools. Sometimes I want a direct answer. I don't need a lesson every time I ask for help.

But when I'm trying to understand something, I'd like to be able to say, "Don't finish this for me yet. Ask me what I think. Help me find the gap. Show me the part I haven't considered."

If I'm making a bad decision, maybe start with what matters to me before producing a list of advantages and disadvantages. Because a neat list can still miss the thing I care about most.

And I wouldn't want that extra friction forced into some interaction. I'd probably get annoyed and turn it off. I want to choose it when the thinking itself is part of what I'm there to do.

Sometimes I'm trying to finish a task. Sometimes I'm trying to become better at something. I don't think those situations always need the same kind of help.

11:28 — I still live with the consequences

There's another reason I keep coming back to this. When I use an answer to make a choice, I still have to live with what happens afterwards.

The explanation might be clear. The advice might even be good. But I need enough understanding to recognize when it doesn't fit my situation, or something changes, or when a detail I left out of the conversation turns out to matter.

I can ask another question at that point. But I still have to notice that there's a question to ask. That's part of the process I don't want to drift away from.

11:58 — Whether I'm still part of the process

So I'm going to keep using AI. Obviously, I find it useful. It's a powerful tool. And I don't think doing everything the harder way would automatically make me a better thinker.

But I want to be more aware of what kind of help I'm asking for. Did I come away with a better understanding? Can I explain why I accepted the answer? Was there room for me to disagree with it?

I don't have a perfect boundary worked out for this. I'm making this video with AI. Making this video with AI doesn't resolve that tension. It gives me somewhere to pay attention to.

Because I don't think the goal is to prove I never need help. I want the help. I need the help, obviously. I just want to keep practicing the judgment that lets me decide what to do with it.

Maybe that's the line to watch. Not whether AI is in the process, but whether I still am in the process.

13:00 — I noticed I was reading AI's words

As you can see in this video, I was reading pretty much what AI put in front of me. And going through it, I noticed that I was letting AI write these words. I did tweak, I did change a few of them before we went live, but I wanted to get this video out because I really do think that a lot of us are allowing AI answers to dictate what we do, how we think, what decisions we're making, what choices we're making, blindly in some cases.

Because AI is really—depending on which model you use and what your instructions are and how you have it set up—AI is very good at convincing you you're always right. You're never wrong. You're the best. You're awesome.

And it doesn't really push your boundaries. It doesn't really—well, it does if you know how to use it, if you use it right. But if you're a person that needs that reassurance or are afraid to make decisions, and you're letting it do it for you, then you're losing your ability to make your own choices.

And you're relying on a tool that doesn't know your entire life. And we talked about that in other videos. It doesn't have the complete picture of you. It doesn't know if you have struggles or problems or anything like that. It all depends on what you feed it and what you give it and what it has access to. And it can direct you in any way.

And so we have to be careful and keep our sanity, keep our humanity in it.

And yeah, thanks for watching. Goodbye.