I’ve been thinking a lot about thinking lately, which is not something I ever thought I’d write. Not in a formal way. More in a quiet, slightly uneasy way that shows up when I’m halfway through a task and realise I’ve reached for ELM almost automatically. Sometimes that feels helpful. Sometimes it makes me pause and wonder whether it’s helping me think or quietly thinking for me. I keep coming across posts and conversations where people say they miss thinking. Not being busy. Not producing outputs. Just thinking. Sitting with a problem. Feeling a bit stuck. Slowly figuring out what they actually want to say. Basically, that uncomfortable moment where you don’t know where to go next. Every time I read something like that, it makes me question things.

I think these questions keep coming up for me because of my background. I studied Artificial Intelligence and Computer Science, so on one level, large language models genuinely excite me. I find them fascinating. I love seeing how far they’ve come and thinking about what we can now build with them. But I’ve also studied psychology, and that part of me is much more cautious, because I have learned to pay attention to habits, cognition, and long-term effects. Not just what a tool can do, but what it slowly changes when we use it every day. Somewhere between those two perspectives is the question I keep coming back to: is ELM helping me think, or is it starting to think for me?

When people worry about AI reducing critical thinking abilities, I don’t think they’re saying humans will suddenly forget how to think. The concern feels much quieter and more long-term than that. It’s about how easy it is to skip the uncomfortable middle part of thinking. That moment where you don’t quite know what you think yet. Where you’re struggling to put something into words. Where things feel slow and messy. At what point do we lose the ability to sit with not knowing? Or the ability to tolerate that discomfort long enough for learning to happen? When does the joy of figuring something out or achieving something start to feel a bit… meh?

LLMs are very good at smoothing all of this over. You ask a question and you get a clean, confident answer back almost instantly. And that makes me wonder: are we slowly losing some of our abilities here? Not just critical thinking, but the ability to sit with our own thoughts, to delay gratification, to stay with something long enough for the longer-term rewards to matter. I’ve noticed this in myself. If I know I can just ask ELM, I’m much less likely to sit with a question for very long. I think that’s a very natural response to being around a convenient tool. But when does convenience start to become a problem?
That’s where the worry comes in. Not that ELM gives answers, but that it can quietly make the process of thinking feel optional.
But at the same time, my job is literally to show students and staff at the university how they can use ELM effectively and responsibly. So, I don’t fully agree with the idea that using LLMs automatically makes us worse thinkers.

More importantly, because of my role, I’ve had to ask myself a different question: how do I convey this nuance to the people attending my training sessions? How do I talk honestly about the concerns, while still encouraging people to use the tool in a way that helps them learn and maintain the skills they actually need?

Trying to balance my own concerns with improving the training experience is what pushed me to experiment more deliberately with how I use ELM. I became quite intentional about it. I try not to start with “do this for me.” More often, I’ll ask things like:

• How could I approach this?
• What should I be thinking about before starting?
• What assumptions might I be making?
• Can you challenge my reasoning here?
• Is there another perspective I should consider?
• I want to discuss a topic. I’ll give you the pros, you give me the cons.

Through this process, I’ve realised that with LLMs, thinking often comes first. Spending time thinking about the task at hand, the process, the output, and what I actually want turns out to be helpful in more ways than one. Not just as a way to exercise your brain, but also to make using ELM more effective, easier, and more intentional. When I use ELM well, most of the thinking happens before I type anything. I pause and ask myself:

• What am I actually trying to do here?
• What am I stuck on?
• What would a good outcome look like?
• What do I already know, and what am I unsure about?

Used this way, ELM feels less like a shortcut and more like a structured way of thinking out loud. I still have to decide what matters. I still have to judge what makes sense and what doesn’t. The responsibility doesn’t disappear. I’ve also noticed that asking for process rather than output keeps me much more mentally present. I have also come across interesting ways of going about a task that I would not have otherwise.

For example, writing this blog. Did I use ELM for it? Yes! But the thoughts are still my own. Earlier when writing a blog, I focused on the outline first. Now, I focus on the content first. What’s in my brain. What am I trying to communicate. What are my thoughts and opinions. Th The structure comes later. The rambling part comes first. When ELM comes in after that stage, it feels supportive rather than substitutive. I’m more confident in the work, and more willing to share it, because I know the thinking didn’t come from the tool. It came from me.
So have we solved the problem of critical thinking in the age of LLMs? Probably not. Not on its own.

We’re already operating in a culture where output is rewarded far more than reflection. The more you produce, the better you’re often perceived to be. Taking time to think about how we do the work has never really fit neatly into that picture, even before AI. If anything, the pressure feels stronger now. Because of that, it would be unrealistic to say that individual habits alone can solve concerns about critical thinking at scale.

That said, I do think how we talk about AI use matters. If tools like ELM are framed purely as time-savers, they’ll be used that way. If we talk about them as tools for reflection, sense-making, and exploring ideas, that opens up a different relationship with them. Tools don’t just help us do things. They quietly shape how we think about doing them.
So, to answer my initial question — is ELM helping me think, or is it starting to think for me?

It’s a bit of both, I suppose. While I actively try to strike that balance for myself, I also try to communicate the same idea in my training. In the end, I think it really comes down to awareness.
Used unreflectively, LLMs can make thinking feel optional. Used thoughtfully, they can help surface assumptions, clarify ideas, and support deeper reflection. For now, I’m trying to stay in that middle space. Curious, but not unquestioning. Open, but not passive.

Maybe ELM can indeed help me think. I just don’t want it to think for me without me noticing.