breme.

Research · Technology and AI

Do Australians trust AI in customer service? What three large surveys show

By BREME · · 5 minute read

In short

Three large surveys of Australians published in 2025 point the same way. Australians are among the most cautious consumers in the world about AI in service: only 23% trust organisations to use AI responsibly, and 58% worry automation will stop them reaching a person. But the evidence does not say customers reject AI. It says they reject AI that fails to resolve, makes them repeat themselves, or blocks the way to a human.

Who this is for: CX, contact centre, digital and operations leaders in organisations planning or already running AI in customer service, from telcos and insurers to health groups, retailers and membership bodies.

The question, and the three studies

Most organisations with a contact centre are now deciding how far to automate service. The business case usually rests on cost. The customer case is harder to see, so we looked at what Australians themselves have said in three large published surveys.

Qualtrics' 2026 Consumer Experience Trends report surveyed more than 20,000 consumers in 14 countries, including 1,502 Australians, and was released in October 2025. Twilio's Decoding Digital Patience study, run with YouGov, surveyed 7,331 adults across seven Asia Pacific markets, including 1,031 Australians, between 28 August and 4 September 2025. Zoom's report, AI alone won't save CX. Resolution will, run by Morning Consult, surveyed 3,509 consumers across seven countries and was reported in August 2025.

All three were commissioned by technology companies that sell service or experience software. That does not make them wrong, but it is worth holding in mind as you read the numbers.

Australians are more cautious than most

Qualtrics found only 23% of Australians trust organisations to use AI responsibly, lower than the worldwide average of 29%. More than half (56%) are concerned about the misuse of data if companies use AI for customer interactions, and 58% worry that automated AI interactions will prevent them from connecting with a human.

Across eight common activities in the Qualtrics study, 51% of Australians preferred in-person meetings or phone conversations over options such as AI assistants or live chat.

Twilio's numbers sit alongside this. 78% of Australians had used AI-powered tools such as chatbots or voice assistants, but only 39% reported satisfaction with them. Nearly half (49%) prefer to start with a human agent even if it takes longer, compared with 42% across the region. Half (50%) said the use of AI in customer service is making them less patient.

Patience tells its own story. Twilio found 53% of Australians see themselves as patient with AI chatbots and 54% with automated voice menus, compared with 88% for live chat and 90% for phone calls with human agents.

What actually frustrates people

The detail matters more than the headline. When Twilio asked about AI-powered service, the top frustrations were:

  • not understanding the question (52%)
  • having to repeat themselves multiple times (48%)
  • scripted or robotic answers (46%)

Resolution, memory and a way out

Zoom's study makes the sharpest point. In its Australian sample, 51% were frustrated by customer service bots that failed to resolve their issues, and 86% said they would likely stop buying from a brand if support did not resolve their concern. The report found resolution, not simply the chance to speak to a human, was the most important driver of loyalty.

The gaps between what Australians expect and what they get are large:

  • 86% expect AI tools to escalate unresolved issues to a human agent; 38% say this happens regularly.
  • 76% expect AI to remember previous interactions; 25% say businesses consistently do this.
  • 74% want AI to anticipate their needs; 28% say it regularly does.
  • 84% (Twilio) say an easy transfer from an AI agent to a human is important.

It costs more than the interaction

Qualtrics found 41% of consumers surveyed said a poor AI interaction led them to cut back on spending with a brand. Globally, Qualtrics reports that people who used AI-powered support were 13 points more likely than average across AI tasks to say they received no benefit from it.

Twilio found that when automated interactions take longer than expected, one in five (20%) stop using the product or switch to a competitor. Australians were more likely than regional peers to downgrade their opinion of a brand after poor service (34% against 29%) and to share the bad experience (32% against 25%). Zoom found 77% of Australians share negative experiences by word of mouth.

There is a second cost that is easier to miss. Qualtrics found 31% of Australians take no action after a bad experience. If an AI channel is quietly failing, many customers will not tell you. They will just spend less or leave.

What the data does and does not show

It does show a consistent pattern across three separate samples: low trust, a strong wish to keep a route to a person, and frustration focused on failure to understand, repetition and dead ends.

It does not show that AI makes service worse in general. These are self-reported attitudes, not measured outcomes, and they mix good and bad deployments. The surveys were run and published in 2025, and attitudes may be moving. Two of the three studies report Australian figures inside a regional or global sample, and none of them, as reported, separates simple tasks from complex or emotional ones. Zoom's data hints at that split: only a minority of Australians ranked direct human contact as their top priority, with speed and agent knowledge rated more highly.

So the honest reading is narrower than either side of the AI debate tends to claim. Customers are not asking for no AI. They are asking not to be trapped.

What we would do with this

Treat AI as a stage in the journey, not a channel on its own. Map where it sits, what it hands over and to whom. Then test it the way a customer meets it:

  • Measure resolution in the AI channel, not containment. A conversation that ends without a fix is not a deflection, it is a delay.
  • Make the route to a person visible and short, and check it works out of hours.
  • Pass the full context at handover so nobody repeats their story. Repetition is the second biggest frustration in the data.
  • Keep AI away from moments that carry money, health, hardship or complaint until you have evidence it resolves them.
  • Listen beyond surveys: read transcripts, watch repeat contacts and track spend after AI interactions, because many unhappy customers will not say so.

Sources

See your journey from the customer's side

We map the end-to-end journey with you, find the moments that matter most, and build a strategy your teams and systems can deliver.

Start a conversation More insights