Engineering Manager's thoughts on AI
I really want to use AI but I also want to keep getting better at my craft.
When I’m reading a document or an email, if it feels AI-generated, I start losing interest. I disengage. I stop caring what’s written.
As an Engineering Manager, this is conflicting. I want my team to use tools that make them work better and faster. But I also feel bad when I secretly think their work is “AI slop.”
When I call it out, they get defensive:
“These are my thoughts… I just used AI to polish them.”
The reality is that most people aren’t great reviewers. They don't know what to look for. They mistake “good grammar” for “good thinking.”
Here is a look at what AI is actually doing to our jobs, where it works, and what could be the future of software engineering.
Why Does it Matter How Something Was Created?
Would you visit an art gallery where all paintings are perfect but created by robots? Would you buy that art?
Even when I can’t tell something is AI-generated, finding out later ruins the magic. It happened to me when I found out that the tiger in the movie Life of Pi was just VFX. It felt like I got cheated.
Turns out, we care about how something was created. Knowing that there were humans behind something, we are in awe. We compare those skills to our own abilities. Which leads to us appreciating it and valuing it. Even if it’s not perfect.
Human mistakes are part of the experience. When you play chess against a computer it just plays at the exact difficulty level you pick. Whereas a human might surprise you with a genius move or a silly mistake. That’s what makes it exciting.
Then Why Are Companies Pushing for AI Adoption?
Actually we don’t care about the creator for every single thing we buy. We’re already buying cars that are built by robots. Because cost matters too.
Companies love processes and systems that can scale. With technology, companies can build more things consistently and predictably. And for cheap.
It happened in manufacturing, farming, and construction. Humans moved up the value chain doing things machines couldn’t do. More white collar jobs.
Until now.
Jobs like software engineering rely heavily on human knowledge (skills, flow and mood), which doesn’t scale very fast. So companies are pushing AI hard. Even if they have to trade away some creativity for speed, they’re ok. More optimization of that will come later.
That’s What’s Happening to the Tech Industry
When tech gets easier, one person is expected to do more jobs. We have seen this before.
The industry created “Full Stack Engineers” and added frontend, backend and databases responsibilities all in one role. Later under the guise of “full service ownership” added infrastructure, testing and security.
With AI, we are seeing more of the same. New terms like “Product Engineer” are popping up. Soon, you won’t just write software, you will build and run the whole product by yourself.
All software engineers will become architects. They will focus on planning, let the AI build the software, then review the work. The future of skills will not be on the middle section of building, but the planning and reviewing steps.
But there is a risk. When the basic code breaks and AI can’t fix it, who will step in? The engineer who can still do that will be in high demand.
How I Use AI as an Engineering Manager
I’m not against using AI. But my relationship with it has evolved.
1. Writing Code
I use it a lot for creating internal tools, dashboards or one-off scripts. Occasionally, I use it to fix small bugs. But I don’t personally write code that goes to external users.
I have built side projects with vibe coding but haven’t really shipped them. AI is great at making things look like they work. It’s terrible at edge cases. It can take your app from 0 to 90 quickly, but the last 10% needs skills I don’t have.
If you are an engineer writing software for real customers, you should care about those skills. You’re 100% responsible for your product, even if AI wrote it.
2. Writing Posts
When I started posting on LinkedIn in 2023, AI wrote my entire first post. It was way better than what I could write back then. But as I learned how to write better, I used AI less.
The same goes for this newsletter. I used to give ChatGPT a few notes and let it write the whole article. Now, I let my own thoughts flow and write it myself.
I use AI to brainstorm, debate and check my post at the very end. I compare my writing to the AI’s writing, sentence by sentence, and pick the best parts.
3. Design Docs
I hate using AI for this. Design docs require deep thinking. You have to know your company and your team. ChatGPT might know what a generic boss wants to see, but it doesn’t know my specific leadership. It doesn’t know what questions my team will ask in the review meeting. You cannot fake that kind of knowledge behind perfect grammar. In fact, using AI too early here paralyzes my thinking and judgment.
4. Brainstorming and Ideation
LLMs are great for bouncing ideas around. It feels like a real conversation, which helps me learn faster. It follows the way I think. My last five chats with ChatGPT were all over the place:
- Crepe business in US
- Design Fishbowl game
- Data pipelines in sports broadcasts
- Vegetarian meal plan
- Seattle to Chicago trip
I treat these chats like talking to a friend. I’ll listen to their opinion, but I don’t trust them blindly but they are not experts. I still do my own research.
5. Summarizing from my chaos notes
I take a lot of messy notes during the week. When it's time for performance reviews or feedback, I rely on AI to help me find the themes in those dates and bullet points. It turns my “brain dump” into a useful summary.
Closing Thoughts
I still feel that “ugh” feeling when I see AI slop. But I’ve realized that my job isn’t to stop myself or my team from using AI. It is to learn and teach how to survive the AI era. So,
Don’t be complacent: Don’t be an AI hater, be an expert who knows how to use the tools.
Give away the boring stuff: Figure out what skills you don’t want to learn (like writing basic boilerplate code) and let AI do it.
Develop your “Taste”: When anyone can make “average” work in three seconds, your superpower is knowing what is truly great.
Think more: Your value isn’t typing words or code anymore. Your value is strategy, judgment and solving hard multi-system problems.
Further Reading:
Matt Schumer’s article “Something big is Happening”
Elena Verna’s article “There’s a short window to get radically ahead by going AI-native. You need to act now.”