Visualised: The AI Job Displacement Timeline (2020–2035)

In 2020, fewer than three in every hundred jobs sat at meaningful risk of automation. By the mid-2030s, that figure is projected to climb past one in three — and the climb won’t be smooth. Drawing on PwC’s three-wave automation framework, this timeline tracks how AI’s reach moves from narrow, structured tasks today into the unstructured, judgement-heavy and eventually physical work of tomorrow. The steepest part of the curve isn’t behind us — it’s still ahead, in the back half of this decade and the first half of the next. Press play to watch the acceleration unfold year by year, and see where your own decade falls on the curve.

The 300 million number and the 30% number are measuring different things

Headlines love to cite Goldman Sachs’ estimate that 300 million jobs worldwide are “exposed” to AI automation. This chart tracks something narrower and more conservative: the share of jobs at high risk of automation, per PwC’s modelling. Exposure means AI can touch some part of a role’s tasks — most office jobs qualify. High risk means the bulk of that role’s tasks are automatable. Conflating the two is the single most common distortion in AI-jobs reporting, and it’s why a “300 million jobs affected” headline and a “30% high-risk by 2035” chart aren’t actually in tension with each other.

We’re still on the gentle part of the curve

Despite several years of AI jobs headlines, the trajectory mapped here puts 2026 only around the mid-teens in percentage terms — meaningfully below the late-2020s and 2030s acceleration still to come. The augmentation wave, which automates repeatable tasks and unstructured information work, doesn’t fully mature until the end of this decade. The autonomy wave, which extends automation into physical and dynamic real-world tasks, doesn’t peak until the mid-2030s. Most of the disruption this chart describes hasn’t happened yet.

2030’s headline number hides who actually bears the cost

The World Economic Forum’s widely cited 2030 projection — 92 million jobs displaced against 170 million created — nets out to a positive 78 million. But a net gain at the global level says nothing about who experiences the loss and who gets the gain. The roles being displaced and the roles being created cluster in different industries, demand different skills, and concentrate in different parts of the world. A net-positive global figure is consistent with a very uneven, very painful transition for specific workers and regions.

Each wave doesn’t replace the last one — it stacks on top of it

The curve bends upward rather than just rising because the three waves are cumulative, not sequential. The algorithm wave (structured data, simple digital tasks) doesn’t switch off when the augmentation wave (repeatable tasks, unstructured data) kicks in — it keeps running underneath it. By the time the autonomy wave adds physical labour and dynamic problem-solving into the mix in the 2030s, all three layers of automation are operating at once. That stacking effect, not any single breakthrough, is what produces the accelerating shape.

Which wave hits you depends on what you do, not when you’re reading this

“2035” isn’t a single deadline — it lands at wildly different times depending on the job. Office and administrative roles are already well into the augmentation wave today. Manufacturing, transport and other physically grounded work largely waits for the autonomy wave in the early-to-mid 2030s. The shape of this chart is a global average; where any one job actually sits on it depends entirely on its task mix — which is exactly what our salary vs. automation risk and task-level breakdown by role charts dig into next.

Methodology

This timeline tracks the share of jobs at high risk of automation, based on PwC’s published three-wave automation framework (“Will Robots Really Steal Our Jobs?”), which estimates roughly 3% in the early 2020s, ~20% by the late 2020s and ~30% by the mid-2030s. Years between these milestones are smoothed for illustration and are not official PwC year-by-year figures. The 2023 and 2030 reference points draw on separate estimates from Goldman Sachs Research and the World Economic Forum’s Future of Jobs Report 2025, shown for context rather than as part of the continuous series, as each organisation uses a different methodology and definition of risk.