There are two dominant narratives about AI and employment. The first: "AI will replace all jobs in 5 years." The second: "AI is a tool, humans will always be necessary." Both are convenient oversimplifications that avoid the more complex, less comfortable truth.
I'm going to try not to fall into either one.
What the 2025-2026 Data Says
Tech employment numbers tell an interesting story. Major Silicon Valley companies significantly reduced junior developer headcount in 2024-2025. GitHub, Salesforce, Google, and Microsoft cut entry-level development roles by 15% to 30% in some departments.
At the same time, senior developer salaries increased in that same period. Demand for engineers who can architect systems, make complex technical decisions, and oversee AI tool output skyrocketed.
That's not "AI replacing humans." That's AI replacing less specialized humans while increasing the value of more specialized ones.
What Jobs Are Actually at Risk
Being specific, the jobs genuinely in structural decline:
- Data entry and document processing: largely automated. Not generative AI — classic automation that simply accelerated.
- Simple, repetitive feature development: boilerplate code, basic CRUDs, simple API integrations. Copilot and similar tools do these in seconds.
- Tier-1 technical support: answers to common questions, basic troubleshooting. 2026 chatbots are genuinely good at this.
- Generic content writing: low-quality SEO articles, standard product descriptions. That market collapsed.
What Jobs Are Growing
The other side of the equation:
- AI systems and prompt engineering: designing how companies integrate AI into their processes requires understanding both business and technology.
- AI systems security: a field that barely existed in 2022 and today has more demand than candidates.
- Complex systems architecture: AI can write code — it can't decide what system to build or why.
- Human-AI experience design: UX/UI for products that integrate AI in ways that feel natural. This is harder than it sounds.
The Problem with the "Co-Pilot" Narrative
The popular narrative says AI is a "co-pilot" that makes you more productive and that's it. The problem is that increased productivity has direct consequences on headcount.
If a developer with Copilot can do the work of three developers without Copilot, companies don't hire three times more developers. They hire fewer and expect more. That's real pressure on employment, especially in junior positions.
I'm not saying it's bad that AI exists. I'm saying minimizing this effect with "but it creates new jobs too" is an incomplete answer for someone who is job hunting today.
What I Would Do If I Were Starting Now
This is what most people ask me. Honestly:
I wouldn't avoid learning to code. Computational reasoning, the ability to understand complex systems, knowing how to read and evaluate code — those skills remain valuable.
I'd focus on layers AI can't easily automate: understanding the business behind the software, communicating well with non-technical stakeholders, making architecture decisions under uncertainty.
I'd learn to use AI as a central tool, not an optional assistant. The developer who ignores AI tools in 2026 is equivalent to the one who ignored Git in 2010.
Conclusion Without Decoration
AI is replacing some jobs. It's also creating others. The net effect on total employment is genuinely uncertain and anyone who tells you they know exactly how it will play out is lying or selling something.
What is certain: the skills that make a developer valuable are changing. Adapting to that change is not optional.