Work with AI so it raises your value

Company AI policy

Also called acceptable use policy, approved AI tools

What

Your employer's rules on which AI tools you may use and what data may go into them.

Why

Pasting code, customer data or secrets into an unapproved tool can breach your contract and is a fast way to lose a job.

AI adoption at work

What

Using AI tools in your daily work as a normal part of how you build, test and review.

Why

Most developers now use them, and some employers treat adoption as a selection criterion. Not using them is a risk in itself.

Example

In the 2025 Stack Overflow survey, 84% of developers used or planned to use AI tools.

Evidence on AI productivity

Also called METR study, AI speed-up

What

Studies that measure how much AI tools really change developer speed, as opposed to how fast people feel.

Why

In a 2025 METR trial, experienced developers took 19% longer with AI while believing they were faster. Measure your own work instead of trusting headlines.

Deskilling

What

Losing a skill because a tool always does it for you.

Why

Reading, debugging and designing are the skills that make you valuable when AI writes more code, and they fade without practice.

Judgement skills

What

Skills that decide whether code and designs are right: critical review, debugging across systems, performance reasoning and choosing what not to build.

Why

As code gets cheaper to produce, judging it becomes the scarce part of the job.

Building with AI

Also called AI-adjacent work, AI features

What

Engineering that puts models into products: model calls, retrieval over company data, agents that call tools, and the cost, security and reliability around them.

Why

Work AI does is shrinking; work that builds with AI is where much of the new hiring is.

Evals

Also called AI evaluations, evaluation

What

Tests that check whether an AI feature gives good answers, using a set of example inputs and a way to score the outputs.

Why

Teams cannot ship AI features safely without them, and engineers who can build them are in demand.

T-shaped engineer

Also called T-shaped, depth and range

What

An engineer with deep skill in one area and working knowledge of several around it.

Why

Depth makes you hard to replace; range lets you move when one area shrinks.

Learning system

What

A small, regular routine for learning: daily use, a weekly read, a monthly experiment, a quarterly deeper project.

Why

It keeps you current through busy periods, where bursts of panic learning fade within weeks.

Proof of AI skill

What

Evidence that you use AI well: a measured improvement at work, a shared setup, or a small project with an honest evaluation.

Why

Listing a chatbot as a skill says nothing. A result with numbers shows judgement.

Example

Cut code review time by 30% with an AI-assisted check, measured over six weeks.