My thoughts
It’s a tool. An amazing and sometimes scary tool. But it wont replace you if you have domain knowledge, connections in the industry you’re in and most importantly good soft skills (listening skills, time management, leadership capability)
Learn how to network and build connections. Read, in office time, not pissing people off, not asking really dumb questions that can be googled. Actually communicate in person. Benjamin Franklin effect…asking for small favours. Mirroring. Being present, relaxed and having fun.
Understand power games. Don’t outshine the master, say less than necessary, don’t put too much trust in friends, learn how to use enemies, perception vs perspective
Someone still has to prompt the chatbots. They don’t yet have intent. So learn some good tactics around prompting.
Understand the problems from a business sense. Ask yourself why is this task needed. What is it doing. How does it work in the larger context of the product/service you are building. Communicate with your seniors. AI doesn’t have real world common sense.
You still have to understand what the AI is spitting out. You need to write tests and debug.
Invest in system design skills. The context window for these bots are not big enough to take in a whole project.
Understand it’s weaknesses. AI’s “creativity” is derivative—it mixes and matches existing knowledge but lacks personal experience, emotions, and intentional originality. True creativity often comes from intuition, life experiences, and emotional insight, which AI lacks.
Don’t just blindly use AI. Think and understand the problem first. Bias towards writing out your own solutions. Think of it like a fancy SO. This will keep your brain sharp and not lead to the total erosion of your skills and understanding of the codebase.
Here follows some Chatgpt answers on the subject:
1. Use AI as a Mentor, Not a Crutch
Instead of asking AI to “write this function for me,” use it to explain concepts, suggest optimizations, or debug code you wrote yourself. This way, you’re leveraging AI without fully outsourcing the thought process.
Example:
- Instead of: “Generate a C++ function to parse this file format,”
- Try: “Can you explain the best approach to efficiently parse this file format?”
- Then attempt it yourself and come back to AI for feedback.
2. Slow Down and Implement Before Checking AI
When you’re under pressure, the instinct is to get the solution ASAP. Instead:
- Sketch out your logic on paper or in comments.
- Write a rough implementation (even if inefficient).
- Then check against AI suggestions to compare approaches.
This builds real understanding and helps you retain knowledge.
3. Use AI to Review, Not Generate
Instead of asking for code, write your own first, then ask AI for a review:
- “How can I optimize this function?”
- “Are there edge cases I missed?”
- “Can you explain what’s inefficient in my approach?”
This shifts AI from being a replacement to being a teammate that helps you level up.
4. Study the AI Output Actively
If you must use AI for code generation:
- Don’t copy-paste immediately—first, try to reconstruct the logic yourself.
- Ask ‘why’ about the decisions in AI’s code (e.g., why a certain algorithm was chosen).
- Modify it slightly before submitting—this forces you to engage with it.
5. Set a Personal AI Quota
Give yourself a daily AI budget—e.g., “I can only generate code snippets twice per day, everything else must be written by me.” This enforces discipline while still letting you use AI when needed.
6. Communicate with Senior Engineers
One reason they might resent AI use is if they feel you’re skipping the learning process. Instead of hiding AI use:
- Be transparent: “I checked my approach with AI, but I wanted to validate with you.”
- Show you understand the code, not just that AI wrote it.
- Ask questions that show deeper thinking (e.g., “I used this pattern—does it align with our team’s best practices?”).
7. Track Your Own Learning
At the end of each week, write down what you learned from AI-assisted coding:
- What patterns did you notice?
- What algorithms/data structures did you see used repeatedly?
- Where did AI surprise you?
This reflection ensures you’re absorbing knowledge instead of just shipping code.
The Key Mindset Shift
Think of AI like Stack Overflow—a great assistant, but if you copy blindly, you’ll stagnate. Instead, treat AI as a collaborator that helps refine your work, not do it for you.
Over time, as you build stronger coding habits, you’ll naturally rely on AI less while still leveraging its advantages smartly. 🚀