AI could replace nearly 12% of jobs - today

For Australian HR professionals, artificial intelligence is no longer a distant buzzword drifting over from Silicon Valley. It now comes with a hard number – and it is big enough to matter.

New research from MIT’s Project Iceberg finds that today’s AI systems are already capable of performing tasks worth 11.7% of the United States labour market, or about US$1.2 trillion in annual wages. The study models roughly 151 million workers and concludes that AI could, in principle, perform work equivalent to nearly 18 million full‑time jobs’ worth of activity. 

Australia’s economy is smaller and more services‑heavy, but structurally similar: finance, professional services, health care and public administration dominate white‑collar employment. If AI can already do this much work in the US, Australian employers and HR leaders should assume the same capabilities are making their way into local offices – if they are not there already.

What exactly is the Iceberg Index?

Project Iceberg is not just another opinionated forecast. It is a large‑scale simulation of the US workforce: 151 million “agents” divided into 923 occupations, tagged with more than 32,000 skills and spread across 3,000‑plus counties. 

The key metric, the Iceberg Index, measures technical exposure: the share of an occupation’s wage bill that comes from skills AI has already demonstrated it can perform at a usable level. The authors are explicit that the Index “captures technical exposure, where AI can perform occupational tasks, not displacement outcomes or adoption timelines.” 

In plain terms, the report is not predicting how many people will be sacked. It is telling employers where AI could take over large chunks of work today if management chose to adopt and integrate the tools.

For HR, that distinction is critical. The 11.7% figure is less a prophecy than a menu: it shows where organisations now have choices about whether humans or machines will do particular tasks.

The visible cuts are just the 'surface”'

Thus far, public debate has fixated on headline tech layoffs and AI coding tools. Iceberg’s message is that this is only the visible tip.

When the researchers look solely at existing AI adoption in computing and technology occupations, they find exposure of about 2.2% of total wage value, or roughly US$211 billion. The report calls this the “Surface Index” and stresses that it represents “only the tip of the iceberg.” 

Below the waterline, the picture changes. Technical capability “extends far below the surface through cognitive automation spanning administrative, financial, and professional services,” accounting in total for 11.7% of wage value – about US$1.2 trillion. 

In other words, AI’s quiet strength is in routine white‑collar work: financial analysis, document processing, back‑office administration, reporting and coordination. One analysis of the study notes that high Iceberg values in several US states are driven by “cognitive work – financial analysis, administrative coordination, and professional services” supporting larger industries. 

Australian employers have very similar job structures in banks, super funds, law and accounting firms, healthcare administration, logistics, universities and the public sector. The tasks Iceberg flags as technically automatable in the US are precisely the tasks that fill many Australian job descriptions.

Which jobs and skills are most easily replaced?

The report does not publish a simple hit list of doomed job titles, but the pattern is clear.

The clearest near‑term targets are structured, rules‑based tasks that are already digital:

• Document processing and data extraction in financial services and insurance

• Back‑office administrative work in health, education and government

• Routine financial analysis, reconciliation and reporting

• Workflow coordination and scheduling across large organisations

The researchers point out that financial institutions now deploy AI for “document processing and analytical support,” while healthcare systems are automating “administrative tasks,” freeing clinical staff for patient care. 

These capabilities map directly onto the day‑to‑day work of claims officers, payroll and billing staff, junior analysts, paralegals, coordinators, contact‑centre staff and many early‑career “assistant” or “associate” roles.

Again, Iceberg measures tasks, not whole jobs. Many roles will be partially automated: AI handles the standardised, high‑volume work while humans deal with exceptions, client relationships and complex judgement. From an employee’s perspective, however, that can still feel like the floor shifting under their feet.

Why Australian HR cannot leave this to IT

Project Iceberg was built to help governments identify “exposure hotspots, prioritize training and infrastructure investments, and test interventions before committing billions to implementation.” It is designed for policy‑makers who want to act before disruption shows up in unemployment statistics.

Australian HR leaders should view it the same way: as a tool to stress‑test people strategy.

Three implications stand out.

1. Exposure is granular – down to tasks and regions: The model spans “151 million workers, 32,000 skills, and 3,000 counties,” offering a detailed picture of how AI interacts with the labour market. Traditional metrics such as GDP, income and unemployment explain “less than 5% of this skills-based variation,” underscoring why new indices are needed. That should be a warning to Australian employers who rely solely on high‑level headcount and turnover figures: exposure will differ not just by industry, but by specific task mix in each team.

2. The productivity race will not be confined to Silicon Valley: The report’s authors emphasise that AI’s impact depends on how quickly organisations package capabilities into tools, integrate them into workflows and encourage adoption. Australian firms competing globally – from banking to business services and higher education — are unlikely to ignore an opportunity to automate 10‑plus per cent of work. The question for HR is whether those gains will be used to support growth and better jobs, or simply to cut labour costs.

3. HR’s legacy will be how exposed workers are treated: Iceberg does not dictate which jobs should go; it shows where leaders have room to choose. That places HR squarely in the frame. Will automation savings be reinvested in upskilling and internal mobility? Or will employees discover their jobs were highly exposed only when a restructure is announced?

Reading Iceberg through an Australian lens

Iceberg is built on US data, but its underlying question is universal: where do current AI systems overlap with what people are paid to do?

For Australian organisations operating under the Fair Work system, that question is wrapped in additional layers – awards, enterprise agreements, consultation obligations and a tight talent market in key specialties. That does not reduce exposure; if anything, it raises the stakes. HR leaders will need to align AI plans with industrial relations obligations and genuine consultation, not simply bolt automation onto old workforce models.

Practical steps for Australian HR teams include:

• Map task‑level exposure. Go beyond job titles. Break key roles into the actual tasks performed and identify which match Iceberg‑style capabilities: document processing, standard analysis, routine compliance checks, basic drafting. Even rough estimates are better than flying blind.

• Create an AI transition framework. Before large pilots scale, negotiate clear principles with executives: redeployment priorities, retraining budgets per exposed FTE, timeframes for role redesign, and how changes will be communicated to employees and unions.

• Shift from jobs to skills. Iceberg is a “skills‑centered metric” by design. Australian employers can mirror this by building internal skills inventories, funding targeted learning, and enabling staff to move from high‑exposure work into growth areas such as client advisory, complex case management, change and implementation.

• Update what reaches the board. Boards and executive teams should not only see vacancy rates and wage bills, but also a simple AI‑era indicator: the share of labour spend tied to highly exposed tasks. In a country where productivity debates are front page news, that number will concentrate minds.

The MIT team behind Iceberg puts it bluntly: “The window to treat AI as a distant future issue is closing.” For Australian HR leaders, that window is narrower still, squeezed between global competitive pressures and domestic expectations around decent work and fair treatment.

AI is already capable of doing a non‑trivial share of the work Australians are paid to perform. Whether that becomes a story of managed transition or unnecessary dislocation will depend less on the algorithms and more on the choices employers — and their HR teams — make now, before the iceberg properly breaks the surface.