Will AI replace programmers?

It changes the work more than removes it. Writing code gets faster and cheaper, so value shifts toward architecture, product judgment and system accountability. The developer who uses AI well outputs like several; the one who ignores it falls behind.

Is it still worth learning to code?

More than ever, but with a different focus: understanding systems, designing, knowing what to ask and how to review what AI produces. Coding shifts from typing to directing and deciding.

Which tech profiles win and which lose?

Those who move up to design, architecture and product win. Purely task-executing profiles doing repeatable work suffer more —exactly what AI absorbs first.

AI by sector

AI in Technology and software

Technology is both the sector building AI and one of the most transformed by it. Writing and reviewing code, documenting, testing and exploring solutions overlaps heavily with what models already do. But a system's architecture, product judgment and accountability for what ships stay yours. AI doesn't remove the one who codes: it changes where their value sits.

Take the test · 2 min The book: the full compass

What's already happening

Code generation and autocomplete (IDE copilots).
Assisted code review and bug detection.
Test and documentation generation from code.
Agents solving tasks end-to-end under supervision.
Faster design, prototyping and option exploration.

Where the line is

AI already does

Write and autocomplete code
Generate tests and documentation
Propose fixes for a bug
Refactor and translate between languages

Stays human

Designing the system architecture
Product and prioritisation judgment
Accountability for what ships
Understanding the user's real problem

—Key occupations

Open each occupation to see its exposure, what changes and what to do.

What to do: the 3 A's

Automate the routine

Delegate boilerplate code, scaffolding tests, documentation and mechanical tasks. It's where AI is strongest and where you add least as a professional.

Augment your judgment

Use AI as a pair programmer: to explore approaches, understand unfamiliar code fast and go further in less time —always reviewing, because accountability for the system stays yours.

Anticipate what's next

Move up to architecture, product design and technical judgment. When writing code is cheap, the scarce —and valuable— thing is knowing what to build, why, and owning it.

The number

Technology occupations score between 69 and 78 out of 100 on AI exposure (mean ~74): highly exposed, and also those that gain the most productivity using it well.

Own aggregation over the AIOE index (Felten et al., 2021) and “GPTs are GPTs” (Eloundou et al., 2024).

Frequently asked questions

Will AI replace programmers?: It changes the work more than removes it. Writing code gets faster and cheaper, so value shifts toward architecture, product judgment and system accountability. The developer who uses AI well outputs like several; the one who ignores it falls behind.
Is it still worth learning to code?: More than ever, but with a different focus: understanding systems, designing, knowing what to ask and how to review what AI produces. Coding shifts from typing to directing and deciding.
Which tech profiles win and which lose?: Those who move up to design, architecture and product win. Purely task-executing profiles doing repeatable work suffer more —exactly what AI absorbs first.

Your next step

Take the free test

Find your personal AI exposure and a concrete next step in 2 minutes.

Start the test →

Get the book

The full compass for living and working in the age of AI, with the 3 A's framework.

See the book →

Create your account, save your exposure profile and access the platform.

Create account →

See all sectors