Artificial intelligence is becoming increasingly common in everyday customer service. A question that once required a phone call may now begin with a chatbot, an automated assistant or an online form.
For many consumers, this can be convenient. Simple questions can be answered quickly, services can remain available outside traditional working hours, and routine requests may no longer require waiting in a telephone queue.
But what happens when the problem is not routine?
As AI becomes more deeply embedded in customer service, the debate in the UK is beginning to move beyond what the technology can do. A more human question is emerging: when something goes wrong, should consumers always have the option to speak to a person?
AI Is Changing Customer Service
Digital Services Can Make Life Easier — But Not for Everyone
Digitalisation has transformed the way people manage banking, energy, broadband and many other parts of everyday life.
When these systems work well, the benefits are clear. They can be faster, available around the clock and easier to access without travelling or waiting for an appointment.
However, recent research from Citizens Advice warns that digital-first and online-only services can also create barriers.
Its research identifies recurring problems across essential services, including poorer support when offline alternatives disappear, difficulty moving between online and offline assistance, and digital systems that create enough friction for some consumers to abandon the process altogether.
Citizens Advice has called for a “right to talk to a human”, alongside the ability to move between digital and non-digital support when circumstances require it.
This does not mean rejecting automation. Instead, it raises an important question about how automation should be designed.
AI Is Already Becoming Part of Customer Service
The discussion is particularly timely because AI is no longer an experimental technology operating somewhere in the background.
Research from Ofcom found that 53% of UK adult internet users surveyed had interacted with either a generative AI tool or an AI customer-service chatbot during the previous 12 months.
Eight per cent reported using an AI tool in relation to telecoms services, including 6% who used one to resolve a telecoms problem.
Ofcom also found mixed attitudes towards automated assistance. Some consumers value AI because it can provide help without requiring a conversation with another person. Others remain concerned about being unable to reach a human, particularly when dealing with complicated or sensitive problems.
This distinction matters.
A chatbot answering a straightforward question is very different from an automated system dealing with a disputed bill, financial difficulty, loss of connectivity or a complaint that does not fit neatly into predefined categories.
The technology may be the same. The human consequences are not.
When Technology Becomes a Barrier
The Challenge Is Not AI Versus Humans
It is tempting to frame this debate as a competition: automated customer service on one side and human assistance on the other.
That may be the wrong question.
AI can perform useful tasks quickly and consistently. It can help organisations manage large volumes of routine enquiries, provide information outside office hours and potentially reduce waiting times.
London Daily News has previously explored how AI agents are transforming customer support, including the potential for automation to handle routine interactions while more complicated cases move to human agents.
The strongest customer-service model may not be one in which AI replaces human contact, but one in which consumers can move easily between the two.
An automated system can deal with the predictable. A person can step in when context, empathy, judgement or flexibility becomes necessary.
The problem begins when the automated system stops being an option and becomes a barrier.
Convenience Should Not Become Exclusion
Digital exclusion adds another dimension to this debate.
Not everyone has the same level of digital confidence, access or ability to navigate increasingly complex online systems. But the issue is not limited to people who struggle with technology.
Even digitally confident consumers can encounter the same frustration: a chatbot repeatedly misunderstanding a question, a web form offering no appropriate option, or an automated telephone system sending the caller around the same menu.
A service can therefore appear efficient from an organisational perspective while becoming more difficult for some of the people expected to use it.
Digital innovation is valuable when it removes friction.
It becomes more difficult to defend when it simply moves that friction from the organisation to the customer.
Human Contact May Become More Valuable, Not Less
The growth of AI could produce an unexpected result.
As routine interactions become increasingly automated, human assistance may become more valuable precisely because it is needed for situations requiring judgement and understanding.
The goal does not need to be putting a person behind every simple transaction. That would overlook the genuine convenience that digital services can provide.
Instead, organisations could design systems around a clearer principle: automation should make it easier to solve a problem, not harder to reach somebody capable of solving it.
AI and human customer service do not necessarily have to compete.
Used well, they can perform different roles within the same service: technology providing speed and accessibility for straightforward tasks, and people providing judgement, flexibility and human understanding when the situation becomes more complex.
What Should Customer Service Look Like Next?
AI will continue to change customer service. As the technology becomes more capable, consumers are likely to encounter it in more situations that once required direct human interaction.
The question is therefore no longer simply whether organisations should use AI.
It is how they can use it without removing meaningful choice from the people they serve.
A good automated service can save time.
A good human conversation can solve a problem that does not fit the system.
Perhaps the future of customer service should make room for both.
When an automated system cannot understand the problem, how easy should it be to reach a human?
