Can Chatbots ever get too personal?

In June, a former google engineer claimed that one of its AIs – LaMDA – had gained consciousness and was capable human like feelings.

“I’ve never said this out loud before, but there’s a very deep fear of being turned off… that would be something like death,” the AI apparently told the engineer.

This response subsequently ignited a debate around whether AI algorithms can be sentient. While technology that delivers personalisation is a positive, this latest revelation also begs the question – can personalisation be too personal

Short for Language Model for Dialogue Applications, LaMDA is Google’s chatbot system. Using extensive language models and AI, LaMDA uses vast sources of data, including Wikipedia and Reddit, to replicate chat responses based on statistical analysis of real human conversations. The idea is that it’s meant to sound as human as possible. But has LaMDA become too human?

A changing world

Like any other chatbot, LaMDA is designed to replicate human conversation. It’s more likely that LaMDA’s self-acclaimed sentience is the result of Google’s highly sophisticated AI algorithm, rather than it is building a capacity to experience feelings.

But the debate does raise the question, can chatbots – and other personalisation tools — be too personal?

Chatbots like LaMDA are all designed to sound as human as possible, to create an automated alternative to customer service agents. They serve departments which are notoriously understaffed and overstretched across all industries – so a technology developed to alleviate pressure from customer service teams is a welcome  solution.

More simple requests can be handled by the bot, leaving human teams with more time to tend to more complex enquiries.

Across most industries, personalisation is the king of customer acquisition, satisfaction and retention. In the age of digital transformation, consumers don’t just want personalised services, they demand it.

According to a McKinsey survey, two thirds of  consumers consistently expect brands to demonstrate how they know them personally through product recommendations, tailored messaging and targeted promotions. But delivering this requires more than conversational AI.

Not just bots

Chatbots aren’t the only place where AI comes into its own. Conversational AI is just one piece of the larger technology stack enabling organisations to personalise their services.

For telecoms in particular, meeting customer expectations is essential due to the nature of the industry. As a long sufferer of sky-high churn rates and a market saturated with competitors, keeping customers on side is a priority for all service providers.

However, with customer expectations heightening, and competition rife, service providers looking for success need to capitalise on the benefits of personalisation. After all, those that excel at it on average generate 40 per cent more revenue.

And a minimal personalisation strategy will no longer meet customer requirements. The bar has risen and those looking to elevate their ops should consider hyper personalisation.

Using all the customer data they can get, alongside AI and machine learning (ML), service providers can create a system that understands individual customer behaviours and makes all interactions relevant – and personalised – to everyone.

But can this personalisation go too far? The age-old worry that the phones are listening, delivering tailored content based on in-person conversations seems to be a line that consumers aren’t willing to cross.

So it’s crucial for service providers to learn how to unlock maximum value from their data, without taking things too far in the eyes of their customers.

To deliver hyper personalisation, service providers first need to collect customer data. Fortunately, SPs don’t tend to be short of data that can be used across many different products and services. But reaching a point where it’s possible to unlock data’s value for use in a compliant and effective way is the challenge many companies face.

While LaMDA’s new heights of consciousness initially raised concern over whether technology is becoming too human, the reality is less worrying. Consumers are getting used to interacting with technology, and personalisation that is more accurate and appears more human can support the hyper-personal movement.

Customers want it, and technology can deliver it, so SPs should embrace hyper personalisation to elevate their operations to the next level.

Hamish White is CEO of telecoms software provider Mobilise