Will AI really replace us (part 2)
Although it’s impossible to definitively predict the future, Generative AI is developing so quickly that it seems to be one of those technologies where the impact will be bigger than we can currently imagine. This is a problem for leaders in customer service, an industry that typically works best when demand is predictable and outcomes are repeatable.
Last week I took a look at what happened over the last decade: digital enabled the automation of many customer service transactions, but this was offset by many new types of human interactions which created more value for companies.
This week I want to lay out some of the scenarios that might lie ahead as AI adoption increases. Note that these scenarios are not all mutually exclusive, it’s possible that several of them could occur over time. (all opinions definitely my own from now on)
1. AI substantially reduces manual workload
This is the headline scenario: AI will fully handle a substantial quantity of customer service interactions, and will improve the efficiency of the remaining interactions being handled by humans. Some companies are already stating an ambition of 70-80% productivity improvement on the workload they see today.
Likelihood: high. Companies are already reporting productivity improvements in the 20-30% range, but the bigger impacts may be more like 5-10 years away
2. AI misses the target compared to the promise
A more pessimistic outlook would see AI fail to meet current expectations. The tools are too unpredictable and fail to win the trust of customers. There are still some efficiency savings, but overall productivity improvements are more in the 30-40% range.
Likelihood: low. It is easy to point to historic examples of chatbots that have failed to meet expectations. Many of the barriers that have prevented digital self-service tools from working effectively in the past will still need to be solved for Generative AI: availability of good quality data, integration of systems to perform actions, intuitive and easy to access user interfaces etc. But it’s worth considering that global investment in AI is around fifteen times what it was ten years ago and many brilliant minds are engaged in making sure it works effectively. Generative AI has made investment in customer service into a strategic priority for many businesses, so this is a once in a generation opportunity to build tools that really live up to expectations.
3. AI usage is limited by compute costs
Companies see great benefits in pilot environments, but are not able to fully scale the solution because of punitive consumption costs of LLMs. This trade-off is especially relevant in the case of offshore operations where labour costs are relatively low.
Likelihood: moderate. Lots of factors at play here. LLMs are getting cheaper all the time at the moment, but pricing is likely to be driven by a push to win market share and not indicative of the underlying economics (take a look at the price next time you book an Uber for a good example of how platform business pricing changes over time!). The biggest factor in reducing compute costs will be smart solution designs that make use of a range of LLMs of different sizes and limit use of the more expensive models to situations where they make a significant difference in quality.
4. Customer interaction volumes increase substantially
AI will make customer service easier to access. Instead of having to call a contact centre or log into an app and search for an answer, customers will be able to ask a chatbot or voice assistant to do it for them. Less friction means more contacts from customers.
Likelihood: high. Whenever we make customer service easier to access, the volume of customer requests increases. A few years ago I worked with a European bank that had introduced a chat channel. They found that for every 10 new chat interactions, call volumes only fell by 6. Most of these additional contacts will substitute for other self-serve channels, like reading an FAQ or changing account details in an app, and so will intensify the impact of scenario (3), driving rising compute costs per customer interaction.
5. Human resources retained for resilience
As the number of frontline Customer Service Reps reduces, companies will become more vulnerable to spikes in demand, with less flexibility to add capacity in the short term. Organisations will need to get more creative about how to deliver sufficient human resource in the moments that matter, which may mean Customer Service Representatives taking on a different substantive role, directly serving customers for only part of the time.
Likelihood: high in regulated industries like Financial Services and Healthcare. Companies in other industries may take more chances but nobody will want to be the first organisation to hit the headlines due to an extended outage of customer service AI.
6. Companies invest in uniquely human capabilities for growth
AI will become commoditised such that AI-driven customer service and operations in different organisations will be virtually indistinguishable. The only way companies will be able to differentiate will be through uniquely human traits like empathy, ingenuity and charm to win customers over.
Likelihood: moderate. A good chance we will see some innovations in human-led customer service for premium brands, but many mass-market businesses will be happy with AI service that is efficient and low-effort for customers.
In summary, it’s inevitable that the number of frontline customer service roles as we see them today will fall substantially, but this is likely to happen over the course of many years. There is time for workforces to prepare for the new skills and capabilities needed to run the AI and for completely new business models that will arise from AI.
Next week I’ll share some thoughts on what I think businesses, employees and public authorities should do to prepare.
Recommended news articles
Vodafone sub-brand MOXI becomes the first UK telco to launch an LLM-based chatbot
What can The Wonka Experience teach us about customer experience?
Latest perspectives from BCG
The Next Frontier in Experience Design
Frontier technologies, including generative AI (GenAI), spatial computing, 5G, and augmented reality, are rapidly transforming our world
These innovations are not just gaining ground individually. They’re amplifying one another—and this technological convergence is powering a revolution in customer experience (CX)
As customer demands intensify, it will be critical for companies to understand how users embrace these evolving technologies