What the surveys say
The honest picture first. Surveys consistently find customers are wary of AI in customer service:
- 64% of customers said they'd prefer companies didn't use AI for customer service, and 60% worried it would make it harder to reach a person (Gartner, 5,728 customers, published July 2024).
- 87% said it's essential to have an option to reach a person when a company uses AI, and 50% said interactions are easier when companies use it (Gartner, 3,566 customers, published August 2026).
- 78% of New Zealanders surveyed want to know when they're dealing with an AI, and half are more likely to use one if there's a clear way to escalate (Salesforce, 500+ New Zealanders, 2024).
- In New Zealand, 62% of people are concerned about businesses using AI to make decisions about them with their personal information (Privacy Commissioner's annual survey, 2025).
Read those together and a pattern shows up. People aren't objecting to a voice picking up the phone. They're objecting to being trapped by it. That's why the second Gartner finding matters as much as the first.
What callers actually complain about
We read 257 Reddit threads where business owners and customers talk about AI phone answering, including New Zealand ones. The complaints are specific:
- No way through to a person. A Wellington patient switched dentists because the AI wouldn't put them through. Another caller took five attempts to get transferred at a restaurant.
- Feeling tricked. The angriest posts come from people who realised partway through that the friendly voice was a bot.
- Being misheard. Accents, place names and background noise, then having to repeat themselves.
- Nothing happening afterwards. A message taken but never passed on, or a booking confirmed that didn't exist. One owner put it well: the AI is 20% of the problem, and 80% is what happens after the call.
What callers are fine with
- Being answered straight away instead of waiting on hold or hitting voicemail.
- An AI that repeats their request back correctly and gets it done.
- Simple, clear calls like booking a time, checking opening hours or reporting a burst pipe.
- Being told it's an AI at the start, with a person available if they want one.
Several business owners say complaints dropped once their greeting said up front that callers were talking to an automated assistant. That's their experience rather than a study, but it matches what the surveys say people want.
The honest trade-off on telling people it's an AI
Research isn't all one way. A 2019 field experiment with more than 6,000 customers found that announcing a voice bot before a sales call made customers much less likely to buy, even though the bot performed as well as good staff (Luo and colleagues, Marketing Science). A 2022 study found that disclosure can lower trust at first, but when the AI fails to solve the problem, having been upfront actually helps (Mozafari and colleagues, Journal of Service Management).
For a business answering its own customers' calls, we come down firmly on telling people. Most callers say they want to know, many can't tell AI voices from real ones, and being found out later does far more damage than saying so at the start. Overseas, the EU now requires it: its AI Act says people must be told when they're interacting with an AI, from August 2026.
Does age matter?
Older people tell surveyors they prefer a human more than younger people do. Some business owners, though, find older callers the most relaxed, treating it like an answering machine that talks back, while younger callers hang up faster. The safe approach works for everyone: speak clearly, keep sentences short, and offer a person early.
How to set it up so customers don't mind
- Say it's an AI assistant in the first sentence.
- Offer a person in the greeting, and make the transfer work in one step.
- Start with after-hours and overflow calls, so regulars still reach your team during the day.
- Teach it your business names, suburbs and prices, so it doesn't guess.
- Make sure every message and booking reaches someone who acts on it the same day.
- Listen to a handful of calls each week and fix anything that sounds off.
Before you commit to any provider, run the calls in our testing guide, and read what goes wrong with AI receptionists. Or ring our AI receptionist on the demo line and see how it feels as a caller.
Sources
- Gartner, July 2024: 64% of customers would prefer companies didn't use AI for customer service
- Gartner, August 2026: customer service survey of 3,566 customers on GenAI and human options
- Salesforce, 2024: New Zealand consumers' expectations of AI agents
- Office of the Privacy Commissioner, 2025 annual survey on privacy
- Luo, Tong, Fang and Qu (2019), "Machines vs. Humans", Marketing Science
- Mozafari, Weiger and Hammerschmidt (2022), "Trust me, I'm a bot", Journal of Service Management
- EU AI Act, Article 50: transparency obligations