banner ad

Navigating the Shift: Balancing Self-Service Tech and Human Interaction in the Age of AI

Since the birth of the internet we've been grappling with the balance between online self-service and service delivered by real human people.

The historic context

To date, interacting with computers has been fairly restricted. Computers relied on structured input and relatively simply choices. Online self-service has largely relied on menus (for choices) and forms (for structured inputs). These were not always intuitive to use - especially for customers. User Experience (UX) and User Interface (UI) Design developed as complex specialities within digital transformation to try to over come these difficulties.

But for many users, this was still simply too difficult. And so they continued to rely on speaking to real human people.

Call centres grew in response to this demand. Operators either did things for customers in the first place, acting as an organic interface between the customers and the systems they were unable to interact with on their own, or a trouble-shooters helping customers to figure out how to interact with the computers or correct the errors they'd made.

The net effect was that we paid call centre operators to act as intermediaries between customers and computer systems because the computer systems were too difficult for the customers to use themselves.

(I am not ignoring the fact that some customer simply prefer to talk to real people. But I question how much of that preference is driven by the fact that it is too difficult to do things any other way?)

Using call centre operators as intermediaries brought its own challenges. It's obviously more expensive to pay operators to act as intermediaries than not to (all other things being equal). The work itself is often very repetitive and not very engaging, leading to high turnover rates and high recruitment and training costs.

And, for customers, it leads to long call-waiting queues, being passed from one operator to another, and dealing with sometimes under-trained operators.

Attempts to relieve some of the pressure, like Interactive Voice Response (IVR) and chatbots were relatively primitive and often only served to increase customers' frustrations.

Much of this cost and frustration can be attributed to Failure Demand*. That is costs incurred because the customer didn't get what they wanted without it. (*As opposed to Value Demand, which is the demand for what customers actually want.)

So what is changing?

In 2023, the capability balance between online self-service and service delivered by real human people shifted dramatically.

This change is being powered by the development of Large Language Models (LLM) and Transformers (like Generative Pre-trained Transformers or GPTs) coupled with improvements in Natural Language Processing (NLP) as exemplified by OpenAI's ChatGPT, Google's Bard and many others.

Intelligent agents are now able to deliver a much more personal and human-like interaction than ever before.

Chatbots and IVRs are progressing from being frustrating barriers to getting through to a real person to being really useful agents that can solve problems quickly, consistently and efficiently.

They can receive information in natural language, respond in natural language, and carry the context through the multiple interactions which make up a typical conversation.

These conversations can be text-based (typed and read) or, using text-to-voice and voice-to-text converters, they can be auditory (spoken and heard). It will become harder and harder to determine whether the voice on the other end of the line is human or automated (or whether it is a person or a computer responding to your chat or email messages).

Such systems will start to replace many traditional menu and form-based computer interfaces with natural language interfaces. AI in customer service will enable more customers to self-serve more, more easily, more effectively and more pleasantly. But it could also have similar impacts on internal systems and interactions with suppliers, distributors and other business partners.

And as such processes become more automated, it will become easier to ensure they are compliant with both internal policies and external regulations.

This will inevitably start with the very simplest of interactions only, but will gradually spread to more and more complex interactions as the technology improves and proves itself.


Just as the internet, and then mobile devices, changed customers' expectations of convenience, so too will this new age of intelligent agents transform the future of customer interaction.

Businesses that don't respond will soon find they are getting left behind.

It is important to start engaging and experimenting now, even if it is only with relative simple and non-core processes.

In order to reduce and better manage the inevitable risks of change, organisations could, for example:

  1. start experimenting with internal processes such as human resources and IT support, and then distributor and supplier facing processes, before extending this to customer-facing processes.
  2. start experimenting with a "human in the loop", like letting AI prepare email responses (or parts thereof) but still having a human read and amend them (if necessary) before sending them.

How is your business rising to this challenge?

If you need support, reach out to me for a conversation on navigating this evolving landscape.📞

See also:

No comments: