In late 2025, Commonwealth Bank — Australia's largest bank — made a decision that caught the attention of every customer service worker in the country. It replaced 90 call centre staff with an AI chatbot called "Hey CommBank." The bot diverted 2,000 calls per week and slashed wait times by 40%.
Then it reversed the whole thing. CBA admitted the move was an "error" — the AI had actually increased work for human agents, not reduced it. The 45 workers who hadn't yet been let go were offered their jobs back.
It's a story that captures exactly where Australian customer service jobs sit right now: under real pressure from AI, but not as replaceable as the headlines suggest.
The AI risk scores tell a clear story
More than 578,000 Australians work in customer service and front-line contact roles. On our AI risk assessment, which draws on Jobs and Skills Australia's GenAI Capacity Study and task-level automation analysis, these occupations cluster firmly in the high-risk zone:
- Telemarketers: 8.0/10
- Call or contact centre workers: 7.4/10
- General clerks: 7.0/10
- Receptionists: 6.6/10
- Insurance agents: 6.5/10
- Bank workers: 5.8/10
- Call, contact centre and customer service managers: 5.8/10
Compare that to concreters at 1.9 or electricians at 2.8, and the gap is stark. The customer service floor of an office building and the construction site outside it are in completely different AI risk universes.
Why these roles are so exposed
The Jobs and Skills Australia GenAI Capacity Study — the first whole-of-labour-market assessment of AI exposure in Australia — named general clerks, receptionists, and accounting clerks among the five occupation types most likely to lose employment from AI. The International Labour Organisation found that 24% of clerical tasks are highly exposed to generative AI, with another 58% at medium exposure. No other occupation group comes close to that level.
The reason is straightforward. Customer service work is built on language — answering questions, processing requests, following scripts, routing enquiries. These are precisely the tasks that large language models handle well. When a customer asks "Where's my parcel?" or "Can I change my appointment?", AI can resolve that without a human ever being involved.
Gartner projects that 75% of customer service interactions will be powered by AI by the end of 2026. IBM reports that 59% of call centre leaders are already investing in AI to cut costs and improve efficiency. In Australia, more than half of businesses are now using AI tools in customer-facing operations.
What's actually happening in Australian workplaces
The CBA story gets the most attention, but it's part of a broader pattern across Australian customer service.
Optus cut 200-300 jobs in early 2026 as part of an organisation-wide restructure. Its AI chatbot, developed with Google, reduced monthly human-assisted enquiries by 15%. CEO Stephen Rue said human workers "would remain critical" but AI would help them "make better decisions."
Telstra shed 650 positions in 2026 and has said AI will cause its staffing numbers to drop over the next five years.
CBA — even after reversing its initial 90-person cut — signed a multi-year deal with OpenAI to give employees ChatGPT Enterprise access. The bank still employs more than 2,400 people in its contact centres, but the direction is clear.
The Finance Sector Union's national secretary, Julia Angrisano, accused CBA of "dressing up job cuts as innovation" and said employees wanted to "work with technology, not be replaced by it."
Meanwhile, a Snowflake study of 2,050 business and technology leaders found that while half of ANZ organisations had cut jobs due to AI, 74% said AI had actually created new roles as job functions evolved.
The CBA lesson: AI isn't as ready as it looks
CBA's reversal is worth sitting with. Australia's biggest bank, with deep pockets and advanced technology, deployed an AI chatbot to handle routine calls — and within months found it wasn't reducing human workload at all. The calls the bot couldn't handle properly still ended up with humans, often in worse shape than if a person had taken them from the start.
This pattern has played out internationally. Swedish fintech Klarna laid off customer service staff in favour of AI, then rehired human agents when quality dropped. IBM replaced much of its HR division with AI and later brought people back.
The technology handles simple, repetitive queries well. It falls apart when conversations are emotional, ambiguous, or require judgement — which is a significant share of what contact centres actually deal with. Complaints, billing disputes, vulnerable customers, edge cases that don't match the training data.
Professor Greg Bamber from Monash University told SBS News that customer service is the second most affected sector from AI, with 20% of roles at risk. But he drew a distinction between routine interactions and complex service — the latter still needs humans.
Where the jobs are going
The most likely outcome isn't mass unemployment in customer service. It's a split.
The bottom of the ladder is where the pressure is sharpest. Telemarketers — scoring 8.0 on our AI risk scale — are among the most exposed occupations in the entire Australian economy. There are just 5,000 of them left, and that number is only going one way. Outbound call scripts are precisely the kind of predictable, language-based work that AI does best.
General clerks (7.0, 286,600 employed) and receptionists (6.6, 182,600 employed) sit in similar territory. Much of their work — scheduling, filing, answering standard questions, routing enquiries — can be handled by AI tools that are already widely deployed.
But the picture changes for roles that involve complex problem-solving, emotional intelligence, or human judgement. Call, contact centre and customer service managers score 5.8 — still moderate, but well below the front-line roles they supervise. As AI handles the simple stuff, the humans who remain will handle harder, higher-value work.
The Parliament of Australia's research paper on AI's workforce impact found that occupations expected to grow include hospitality workers, nursing professionals, and construction labourers — roles defined by physical presence, empathy, and unpredictable environments. Customer service sits on the other end of that spectrum.
What this means for 578,000 Australians
If you work in customer service, the honest picture looks like this:
High exposure: Outbound telemarketing, basic inbound enquiry handling, data entry, form processing, appointment scheduling. These tasks are being automated now, not in five years.
Lower exposure: Complaints handling, sales conversations, supporting vulnerable customers, relationship management, escalation decisions. These require the human skills that AI consistently struggles with.
Growing demand: AI oversight, chatbot training and quality assurance, complex case management, and roles that blend technology and customer skills. The Snowflake study found that 74% of ANZ businesses said AI had created new roles.
The JSA study noted that proactive planning, critical thinking, and communication skills are "essential" for workers in exposed occupations. The workers who adapt — learning to work alongside AI tools rather than competing with them — are better positioned than those who don't.
Wondering where your role sits? Check your occupation's AI risk score or browse the full rankings to see how customer service compares across all 358 Australian occupations.