Uber last week cut 10 percent of jobs in its customer-service operation as part of a restructuring designed to “embrace artificial intelligence”.
The move is the latest evidence that the contact-centre industry’s long-standing reassurance – that AI will help human agents rather than replace them – is beginning to unravel.
Major employers are increasingly using automated chat and phone systems to handle the everyday requests that once filled agent queues, and fewer people are needed to answer them.
Speaking to UC Today, Ashish Nagar, founder and CEO of customer experience platform Level AI, predicts employment in customer-service roles will fall by 10-30 percent over the next five to 10 years, depending on the sector.
“Once a week I get the question, what happens to CS [contact centre] jobs in three to 10 years?” Nagar said.
“My answer has been: a 10–30 percent decline from current levels over five to 10 years and dependent on verticals.”
That is not a prediction of an overnight collapse in contact-centre employment, but is a clear rejection of the idea that AI will simply make the existing workforce more productive while leaving its size untouched.
Uber Makes the Link Explicit
Uber’s cuts, reported on 22 July, affected 10 percent of its community operations team, which is responsible for customer support.
An Uber spokesperson said the changes were intended to “simplify operations, strengthen in-person collaboration, and continue to embrace AI”.
In an internal memo, Megha Yethadka, Uber’s vice president of global community operations, said: “We cannot scale frontier technology on top of fragmented processes.”
The company has not said AI alone caused every role reduction. But the timing and wording matter. Uber is reducing support capacity while expanding its AI strategy and directing users who seek help through its app towards automated support.
That goes beyond familiar agent-assist uses, such as generating a call summary, finding a knowledge-base article or suggesting a response.
When a chatbot or voice agent resolves an issue without a person entering the interaction, it is not assisting a human agent. It is doing work that an agent would previously have performed.
The Cuts Are Already Happening
Uber is not an isolated example.
Bloomberg has reported that the Commonwealth Bank of Australia has cut hundreds of contractors from a Johannesburg chat-support operation as AI took on more customer conversations. The bank said in May that its chat platform was resolving almost nine in 10 conversations without agent input.
CBA said it added more than 140 roles across its Australian contact centres over the previous six months, but that does not negate the impact on workers at the outsourced provider, Nutun, whose contract with the bank was reduced.
Microsoft has also expanded automation across its support operation.
Its support workforce – a mix of employees and contractors – has reportedly fallen from about 50,000 to 40,000 in recent years.
More complex problems still need human support. But the number of issues that can be solved automatically is expanding, and every new category of automated enquiry reduces the volume of work available to people.
Cost Is the Real Driver
The business case is direct. AI can deal with basic requests around the clock at a lower cost than a large live-agent operation.
That includes booking changes, order tracking, account queries, receipt requests, password resets and basic policy questions: the repetitive requests that make up much of contact-centre demand.
Hyatt Hotels, for example, is using AI to automate reservation modifications and receipt requests. Pat Nestor, who leads the hotel group’s AI and data analytics operation, said cutting support costs was “clearly a driver”.
Hyatt cut 30 percent of its in-house Americas support staff last year, though a spokesperson said that decision was unrelated to its AI deployment.
But if a company automates the same simple work performed by agents, then shrinks the human operation, the workforce consequences of AI are real regardless of the language used to describe them.
Replacement Will Be Gradual, Not Total
Nagar believes the pace of job losses will be constrained by enterprise technology and customer trust.
“The easy stuff was already automated; new capabilities just do it better,” he said.
He also argues that the main obstacle is not always the AI itself.
“It’s the API, not the AI,” Nagar said, referring to the challenge of connecting a voice or chat agent to the many legacy systems needed to complete a transaction.
In healthcare, financial services and insurance, customers may also be unwilling to trust AI with sensitive or high-stakes decisions.
“There are many roles, in which humans fundamentally don’t trust AI.
“If, unfortunately, your child is sick, would you talk to an AI nurse practitioner over the phone to build a treatment plan and trust it?
“I am super tech optimist. The future though, is more like the on-prem to cloud migration, which took 10+ years than the overnight miracle people expect.”
