You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 7 minutes
Key takeaways:
- AI has removed the friction that used to build critical thinking naturally.
- Less confidence in judgment leads to more AI deferral.
- Leaders should make critical thinking an explicit expectation and build it into reviews and postmortems.
Early in my engineering career I led a project. It was one of those formative experiences where I encountered many of the classic tropes of software engineering – unknown unknowns, unruly scope, and even cache invalidation woes. I remember telling my manager it was painful, but despite the slog, those experiences felt like gold.
Skills hard won and lessons (painfully) learned in service of judgment and taste. Engineers encountered these learning opportunities as a natural consequence of how software is made.
However, AI has changed how we build software, and the landscape of learning has changed along with it. As new junior engineers enter the workforce as AI-natives, leaders need to consider how this shift affects a skill we may have previously taken for granted: critical thinking.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
Why critical thinking matters now more than ever
Almost everything we do relies on the ability to independently apply creativity, domain knowledge, and discernment to situations that don’t have straightforward answers. It’s embedded into the whole lifecycle of development from architecture with competing requirements, to translating abstract requests into code, and debugging a new feature that’s flailing in production.
As critical thinking is so fundamental, it’s often invisible until it’s absent: plausible implementation, wrong problem; passing tests, unintended behavior; functionally correct, users aren’t happy.
AI makes lines of code quicker and cheaper than ever, meaning the importance of critical thinking is growing too. Code was never the whole job. It is just one artifact left behind by countless acts of creativity and discernment. A byproduct of critical thinking.
As the barriers to implementation fall, our contribution becomes: what to build, why this way, and how can we tell if AI output is correct? When the thinking step is skipped, the code may not look wrong initially, but the system suffers, and so do users.
Experienced engineers have spent years accumulating the judgment needed to make these calls. Junior engineers haven’t had that same opportunity yet.
What AI changes about how critical thinking develops
Imagine teaching someone to ride a bike. You can write down every step, and explain balance, momentum, and steering. None of that replaces actually being on the bike.
Critical thinking is much the same. We can teach engineers frameworks for breaking down problems, evaluating tradeoffs, and questioning assumptions. Yet there’s a difference between knowing what to do and actually doing it. It’s a skill that develops through practice.
Once upon a time, a junior’s day might have looked like this:
- Struggled to understand unfamiliar code.
- Made a bad assumption.
- Shipped something that broke.
- Debugged it.
- Explained your reasoning to a senior engineer.
- Realized you had misunderstood the problem.
- Tried again.
Over time, those efforts, the rounds of trial and error, consolidated into the foundations of critical thinking.
AI-assisted development can speed up access to knowledge and remove unproductive friction. That’s a huge advantage, but some of the obstacles smoothed over in service of output are more than inconveniences – they’re opportunities. Some friction is formative.
Over time, there is a risk that thinking is traded off for productivity, and a troublesome feedback loop emerges: the less confidence an engineer has in their own judgment, the easier it is to defer to generated answers. The more cognitive work they defer, the fewer occasions they have to build judgment.
Engineering leaders may already be feeling the early consequences. Plausible implementations are hitting production with unvalidated assumptions that crumble under real user pressure. Output alone can no longer be assumed as a proxy for thinking and learning. Gaps in critical thinking are becoming more visible, and for AI-native juniors, the skill is harder to build organically.
More like this
What can leaders do to encourage critical thinking?
The learning environment has already changed. Leaders will need to adapt too, working intentionally to safeguard the environment that fosters critical thinking.
A junior engineer shouldn’t be expected to arrive with fully developed critical thinking, or be solely responsible for developing it. The team and systems around them should create opportunities to practice, safely and visibly.
Make the ask explicit
Your whole team needs to know that critical thinking is as much a part of the job as shipping PRs. Making this expectation explicit may be particularly important to junior engineers as recent grads may be used to optimizing for completing tasks, and unaware that reasoning is part of work.
Leaders can start by creating a shared understanding of what critical thinking means to your team, and make that definition part of your team’s working norms. You can refer back to it in onboarding, feedback, and reviews. Once you’ve named it, you can make it coachable.
Protect the learning environment
Building critical thinking requires you to expose unfinished thinking, ask naive questions, and change your mind. This calls for a vulnerability that can feel intimidating. You can’t practice creativity without getting the answer wrong from time to time. You can’t practice judgment if it feels like you have to pretend you already have it.
Psychological safety should mean you feel supported to try, and to fail, even if you used AI along the way. When AI can create a full implementation in seconds, there can be undue shame and stigma in admitting you need help. Ask yourself, can engineers on my team openly say “I don’t understand what the AI generated.” Don’t let the use of AI change the stakes for learning in public.
Double down on everyday engineering practices
The best opportunities to build critical thinking are already embedded in standard engineering rituals. Code reviews, RFCs, incident response, and debugging are not just mechanisms for improving the quality of software – they are the “bike riding” of software development.
- PRs that explain not just what changed, but why, including alternatives and tradeoffs.
- Breaking work down enough that assumptions become visible, requiring teams to discuss how the pieces fit together.
- Postmortems that ask not just “what broke?” but “what did we believe that turned out not to be true?”
These aren’t new ideas. They’re an established engineering discipline. What is new is the temptation, and sometimes pressure, to bypass them with automation. The learning value comes from engagement, through active participation, and conscious effort. Not every PR needs a lengthy debate, but leaders can identify interesting ones, and create space for quality discussion.
You don’t need to add layers of heavy process. Instead, protect the moments where your team already thinks together. Make sure juniors have the chance to get on that bike!
Use AI to practice thinking
A senior engineer I worked with would always ask “what would happen if…” on PRs. Over time, I would quiz myself first before asking for a review. Repetition and exposure helped me internalize the habit of testing my own reasoning.
AI can play a similar role if we use it to prompt thinking rather than replace it. Coach junior engineers to use AI for more than being a “human in the loop” – accepting suggestions and approving actions. Instead, keep the cognitive work in the loop. Ask an AI assistant to challenge an assumption, argue an alternative, or interview the engineer about their approach before offering a solution.
A good mentor doesn’t hand you the right answer, they draw it out of you. If we tailor our AI to be a supportive learning partner, we reduce the risk of diluting learning opportunities.
Make progress visible
Juniors need visibility into their progress; where they’re at and what is expected of them. Likewise, leaders need concrete signals to recognize when someone is thriving, stuck, or perhaps becoming overreliant on AI. We can no longer assume growth goes hand in hand with tickets shipped, especially for critical thinking, but we do need something instead.
Critical thinking should be represented in career frameworks, and discussed in 1-1s and performance reviews. To be meaningful, capture it as explicit behaviors that show critical thinking at various stages of progression – be more specific than just “do” critical thinking. Let this become a shared language for managers and their reports to collaborate on goals, celebrate success, and make feedback meaningful and actionable.
From early career “requires significant guidance to identify assumptions” to “independently validates assumptions before implementation.”
Progress doesn’t look like shipping more work, or shipping faster. It might even look like picking up a ticket then advocating for not shipping it at all.
New York • September 8 & 9, 2027
Loved LDX3 New York? Pre-sale tickets for 2027 are now available.
Protect the learning, not the pain
I didn’t realize at the time, but the grind I was pushing through when I ran that project contributed many building blocks to the critical thinking skills I have relied on ever since.
AI-native junior engineers are coming up in a different environment, one where some of the experiences that contributed to critical thinking are being eroded. That skill hasn’t become less important.
Critical thinking has always been foundational, but now that AI can make it easier to bypass, we risk a bigger gap between functional code and a well-considered feature.
We don’t need to recreate the pain of the old learning environment, but when AI removes the pain, leaders need to account for what might have been lost with it to ensure we protect what’s important. Some friction was drag. Some friction was education.
