I asked AI about the future of AI (Yes, AI-ception)
Published 12 days ago
I wanted to know: where is AI going in the next 10+ years? Is diving deeper into large language models (LLMs) in 2025 still a good bet for developers like me? What jobs will rise? What skills will matter?
So I ran a full research sprint — from EU forecasts to WEF reports to expert interviews — to find some real answers. Here’s what I learned.
Before we start: who am I, and why should you trust my take on the future of AI?
I’m Alex, a machine learning engineer with a Master’s in AI and over two years of hands-on experience. Over the past year, I’ve worked directly with large language models (LLMs) — not just using them, but fine-tuning, deploying, and optimizing them at a systems level. I understand how LLMs work under the hood, from tokenization to transformer internals, and I use that knowledge to build real-world AI applications that deliver. This blog is where I share what I’ve learned — and where I think we’re headed.
TL;DR: AI isn’t taking all jobs — It’s changing them
LLMs like GPT-4 aren’t replacing developers, analysts, or even most writers. But they are taking the boring parts of our jobs — the boilerplate code, the report summaries, the SQL cleaning.
If you know how to work with AI, your productivity explodes. If you don’t, you risk becoming obsolete. Simple as that.
This applies everywhere — from backend teams to boardrooms. In 10 years, almost every knowledge job will be shaped by AI. Not replaced. Reshaped.
The rise of the “AI-augmented worker”
Instead of replacing devs, AI helps them.
Think of it like this: LLMs write skeleton code. Humans still design the architecture, fix the edge cases, write the tests, and decide what to build.
That’s not going away.
But here’s the shift: More of your work will involve prompting, reviewing, integrating, and managing LLM outputs.
So if you know how to use these tools well, you’ll do in days what used to take weeks. You’re not coding less. You’re coding smarter.
Routine tasks are dying fast — but complex work is thriving
Across all fields, AI is already taking over repetitive tasks:
- drafting emails
- creating legal templates
- analyzing simple data
If your job is 80% rules and templates, AI is coming for it.
But jobs with creativity, judgment, or messy edge cases? AI helps, but it doesn’t replace.
This includes most engineering roles, most leadership roles, and any job where thinking on your feet matters.
New jobs are already emerging
Entire new categories are forming:
- Prompt Engineers – the ones who know how to "talk" to LLMs
- AI Product Designers – designing how humans interact with AI tools
- Data Curators & Trainers – prepping the training sets and tuning the models
- AI Governance Officers – monitoring models for bias, ethics, and compliance
- AI QA Analysts – testing AI behavior and outputs
Then there’s the “AI + X” roles: AI + medicine, AI + finance, AI + education. These hybrid roles are already hiring — and the trend is just getting started.
So, should you specialize in AI in 2025?
Short answer: yes. Long answer: hell yes.
The AI field isn’t saturated. Demand is way up. Especially in Europe, where digital talent is behind targets and LLM adoption is still heating up.
If you’re a full-stack developer with ML experience, you’re in a great position.
Adding deeper LLM skills — prompt design, model deployment, fine-tuning — can push you into:
- AI architect roles
- AI product roles
- Or even into building your own AI tool
You stop being “a dev who’s good with AI” and become “the person driving the AI part of the product.”
Final Thoughts
The AI wave is real — and it’s not about mass job loss. It’s about job transformation.
Learn how LLMs work. Build with them. Deploy them. Understand their limits. And figure out how to make them work for you — not against you.
This isn’t the end of programming. It’s the start of a new kind.
P.S. You can see the exact question I asked ChatGPT to run this deep research here. P.S.2 While the research paints a balanced picture, I don’t fully agree with it. Personally, I believe that in 10 years, AI will surpass human intelligence in many areas, and companies will rush to replace humans with AI to cut costs. The quality of output might drop — marketing campaigns may get sloppier, for instance — but many of these jobs don’t require perfect precision. Once LLMs gain memory, better hardware access, and control over interfaces (like mouse/keyboard/screen), they’ll be capable of doing most knowledge work. In my view, that shift is coming — fast.