I remember the first time I used ChatGPT. It felt like magic. But if you’re anything like me, that "magic" feeling has worn off a bit, replaced by something more practical and honestly, a little overwhelming.
Everywhere I look, there’s a new tool promising to do my job for me. But as we move deeper into 2025, I’ve been realizing that the conversation is shifting. It's no longer about what the AI can generate; it's about why we are generating it.
Here are a few things I’ve been thinking about regarding where we’re heading.
The End of the "Prompt Engineer"
For a minute there, we thought "Prompt Engineering" was going to be the career of the future. I’m not so sure anymore. Models are getting smarter at understanding intent.
I’m finding that I don't need to be a wizard with my words to get a good result; I just need to be a good director. The friction is disappearing. This is great, but it means the barrier to entry is effectively zero. If anyone can create "good" work, then "good" isn't enough anymore.
Agentic AI: From Chatting to Doing
The biggest shift I'm seeing in my own workflow isn't just generating text or images—it's agency.
We are moving from chatbots that answer questions to "Agents" that perform tasks. I’m starting to use AI to:
- Plan entire sprints, not just write the tickets.
- Refactor codebases, not just write functions.
- Research market trends and summarize them into actionable PDFs.
It’s less of a tool in my hand and more of a junior assistant sitting next to me.
Taste is the New Moat
This is the hill I’m willing to die on: Taste is your most valuable asset.
When AI can generate 1,000 variations of a UI design in 30 seconds, the skill of making the design becomes cheap. But the skill of choosing the right one? That becomes expensive.
I’m focusing less on grinding out pixels and more on curating. Developing an eye for what connects emotionally, what feels authentic, and what feels "off." AI is terrible at nuance. It doesn't know why a certain shade of blue feels nostalgic. It just knows the hex code.
That’s where we come in.
Keeping it Real (Literally)
There’s a concept called "Model Collapse"—the idea that if AI models train on too much AI-generated content, they get weird and glitchy. They need human data to stay grounded.
I think the web is going to go through something similar. We are already drowning in mediocre, SEO-farmed AI articles. I believe 2025 will be the year of the verification backlash. People are craving human connection, imperfect stories, and videos that haven't been smoothed over by an algorithm.
Final Thoughts
I’m optimistic, but cautious. I’m using these tools every day, but I’m trying hard not to let them replace the thinking part of my brain.
If you’re building a portfolio or a product right now, my advice is simple: Don’t just show what you built. Show how you made the decisions. Show the messy sketches. Show the human part.
Because ironically, in the age of AI, being human is the best USP we have.