Every few months a new report predicts that AI will displace some enormous percentage of jobs. Goldman Sachs said 300 million. McKinsey said 400 million by 2030. The headlines are alarming, and they generate a lot of clicks.

But what’s actually happening? After two years of generative AI being widely available, jobs are changing, but the wholesale replacement hasn’t materialised in the way many expected.

The Prediction Problem

Most job displacement predictions share a common flaw: they analyse tasks, not jobs.

A report might determine that 40% of the tasks in a given role can be automated by AI. The headline becomes “40% of jobs at risk.” But that’s not how work operates. If AI handles 40% of your tasks, your job changes — it doesn’t disappear. You do the remaining 60% plus new things that become possible because the routine work is faster.

This is what happened with spreadsheet software, email, and the internet. Each technology automated significant portions of office work. Employment in offices didn’t collapse. The work transformed.

What AI Is Actually Replacing

There are genuine cases of AI reducing headcount. They tend to follow specific patterns.

High-volume, template-based content. Product descriptions, basic news summaries, and routine reports are increasingly AI-generated. Companies that employed teams to write thousands of similar product listings have reduced those teams.

Tier-one customer support. Chatbots have improved enough that many companies have reduced their front-line support teams. The humans who remain handle escalations and complex issues.

Basic data entry and processing. AI that reads documents, extracts information, and populates databases is replacing manual data entry roles. Insurance claims processing, invoice handling, and similar repetitive document work is a clear area of impact.

Some translation work. Routine business translation requires less human involvement now. Literary and specialised translation still needs human expertise, but straightforward document translation is increasingly automated.

What AI Isn’t Replacing (Yet)

Despite predictions, several categories of work have proven more resilient than expected.

Anything requiring physical presence. Trades, healthcare delivery, childcare, hospitality, construction — these require being somewhere and doing something physical. AI can assist with scheduling or diagnostics, but the core work still requires a person.

Complex judgement calls. Legal strategy, medical diagnosis in ambiguous cases, senior management decisions — these involve weighing incomplete information and accepting responsibility for outcomes. AI assists with research, but the judgement remains human.

Relationship-dependent roles. Sales in complex B2B environments, counselling, teaching, mentoring — these depend on trust and rapport. AI tools help with preparation, but the relationship can’t be automated.

Creative work with a point of view. Content with a distinctive voice, genuine perspective, and original insight still requires a human mind. The demand for generic content is falling. The demand for distinctive content isn’t.

The Real Pattern: Augmentation

In most workplaces, the pattern is augmentation rather than replacement. Workers use AI tools to do their existing jobs faster or better.

A marketing team that previously produced four campaigns a month now produces eight with the same headcount. A software team ships features faster because AI assists with code generation. A financial analyst covers more companies because AI handles the initial data gathering. This is productivity growth, not job elimination.

The firms seeing the most benefit from AI are typically those that invest in helping their existing staff work with the tools. The team at Team400 has observed that organisations getting real value from AI almost always pair the technology with structured training and change management — the AI alone doesn’t produce results.

The Transition Costs Are Real

None of this means the situation is painless. Transition costs fall unevenly.

Workers whose roles are genuinely displaced need to find new work, often in different fields. Retraining programs exist but are patchy in quality. The people most affected tend to be those with fewer resources to manage the transition.

Mid-career workers face the hardest adjustments. They have mortgages, families, and specialised skills that may be less valuable than they were five years ago.

Younger workers face a different challenge: the entry-level roles that traditionally provided training are among the most affected. If junior positions shrink because AI handles the basic work, how do people develop the skills needed for senior roles?

What to Do About It

Whether you’re worried about your own role or managing a team through these changes, some practical responses.

Learn to work with AI tools in your field. Not to become an AI expert, but to be competent with the tools becoming standard. The person who uses AI effectively in their domain is more valuable than the person who ignores it.

Focus on judgement, relationships, and context. These are the areas where human involvement remains essential and where your experience matters most.

Build breadth alongside depth. Specialists whose narrow expertise aligns with AI capabilities are more vulnerable than people who combine domain knowledge with communication and adaptability.

Watch your industry, not just headlines. Talk to people in similar roles at different companies. Real examples are more useful than forecasts.

The future of work is neither the catastrophe some predict nor business as usual. It’s a messy, gradual transformation where some roles shrink, others grow, and most simply change. The best preparation is staying informed, staying adaptable, and staying useful.