Ethical AI = Customer Trust
In the last couple of years, I have said the letters “AI” substantially more often than I’ve said the names of my own children, and I don’t know how I feel about that.
It’s not just in work life. Conversations with friends and family inevitably drift into the subject of AI: how we are using it in our personal lives, how it might affect work in future, and whether it’s all a good thing or not.
The common feeling behind most of these conversations is that the big impact of AI is yet to come, and people are generally a bit worried about it. It’s like a tidal wave that is still many miles offshore, but we can hear its distant rumbling. And like a tidal wave, many people feel like it is something that is being done to them, rather than something they have any control over. The only option is to reach higher ground or get washed away.
This is a sign of a big problem. This incredible, magical technology that offers the potential to transform completely how we as humans live our lives, is instead being rolled out as a kind of done deal: this is what you are getting, there is no time to pause and think how best to use it.
Stepping down from the philosophy into the slightly more manageable world of customer service, we see the same thing in microcosm. Customers know that they need to interact with AI to get service, and there will be more AI in future, but they don’t trust that the AI has been built in their best interests. They have had bad experiences of bad AI in the past and they know that many companies will be implementing AI primarily as a cost-saving measure.
In customer service AI, there is a strong case for less haste, more speed. By taking a step back and thinking about what type of service we want to provide - not just trying to fit everything into what we think AI can do - then we stand a better chance of earning customer trust.
Most importantly, any AI deployment should fit into a defined ethical AI framework.
Ethical AI for customer service
An ethical AI framework goes beyond responsible use of AI, to cover how AI affects and influences human work and personal life, as well as society as a whole. For customer service, it could look something like this:
Under each of these four sections are design principles which ensure the ethical and high-trust deployment of AI, for example:
Service Access
Avoid “doom loops” where customers have no access to human support when they need it
Design services that can be used by those with specific accessibility needs
Ensure traceability and recourse if something goes wrong
Design journeys that build customer trust over time (e.g. by ensuring a customer never has to repeat themselves, or provide unnecessary personal information)
Data Usage
Implement controls to prevent customer data being leaked
Mitigate bias in training data and outputs
Assure the accuracy of outputs by testing against ground truths
Avoid obtrusive surveillance of customers
Uphold a customer’s right to be forgotten by AI
Dedicate resource to protect customers from external scams and deepfakes
Employee Experience
Equip Customer Service Representatives (CSRs) with tools and skills to handle the more challenging work, when AI has automated the simpler tasks
Avoid excessive employee surveillance
Protect employee agency and creativity in their work
Control AI bias that could affect hiring, reward and career paths
Support and equip employees to build employability for new in-demand roles
Prevent employees from using unauthorised AI tools to serve customers
Sustainability
Monitor, control and reduce the carbon cost of AI model usage - taking into account that usage increases when services are easily available to customers
Avoid using advanced AI (Generative, Agentic etc.) for situations where lower-compute automation would suffice
Note: this is a non-exhaustive list, and it doesn’t include some ethical AI considerations that are less relevant in a customer service context, like protection of copyrighted intellectual property.
This is a long checklist of tasks which can seem onerous to customer service teams that are already under pressure to deliver results from AI. But they are necessary and worth the effort - when AI meets customers, things only ever go to plan if we have taken the time to earn those customers’ trust.
Recommended news articles
Forbes: AI or the human touch? Striking a balance in customer retention
Knowledge at Wharton: Can AI fix what’s wrong with customer service?
Bloomberg: Call center workers are tired of being mistaken for AI
Latest perspectives from BCG
AI is working, but is it working for everyone?
It’s hard to find a workplace today where AI isn’t part of the conversation or the work itself. From generating content to streamlining decisions, it’s showing up in ways that would have seemed unlikely just a few years ago.
And yet, beneath the surface of this rapid adoption, something more subtle is unfolding—a tension, a gap, a growing sense that while AI is moving fast, not everyone is moving with it.