There are so many organisations having their say on ChatGPT, but not everything that is being said should be taken at face value.

Companies wondering how they will be able to best leverage large language models (LLMs) – whether that be GPT, LaMDA, Galactica, Megatron-Turing or any other model – need to ensure they separate fact, fiction and fantasy.

We’ve been innovating in the conversational AI space for nearly two decades and as technology continues to advance, we are constantly asking ourselves serious questions about how it may, or is, disrupting the value we can add. This drives us to innovate, so that our customers can maximise real business value from our solutions.

The hot topic of LLMs is not new to us. Our team has been exploring and investigating them for about two and half years now.

One of the enterprise-level features of the Gluon release of our V-PersonTM Technology, which we launched on 16 May, is native support for LLMs. Like all technology, the value lies in how it is used, and this is also the case with LLMs.

An example of how we are using LLMs is theme matching. This has come about following many conversations with our customers and partners and listening to their concerns, challenges and requirements. This use of LLMs is value adding; ensuring accuracy of responses in real-time, at a significantly lesser cost and greatly reduced time and effort compared to similar possible methods.

When going down the LLM route, the questions companies need to be asking are:

  • What is the conversational AI provider able to deliver, that can’t simply be done by using LLMs such as ChatGPT directly?
  • What value is the conversational AI provider adding to your business through real customer use cases they are showing?
  • Can the conversational AI provider back up and demonstrate what they are saying?
  • What is the data privacy, security and sovereignty safeguards that the provider has in place?

The reason these questions must be asked is because simply using ChatGPT to do something doesn’t mean it adds any value. Companies need to look for a provider who delivers conversational AI solutions that address real business challenges and are designed to deliver real business value.

Let me give you an example of what I saw recently on a webinar. Presenting an example of a transactional chatbot, the company demonstrated how easy it is to build a currency converter using ChatGPT.  Good so far.  But:

  1. How many of us have been using Google to do that for as long as we can remember? and,
  2. This is not as a transactional chatbot, and to call it one is disingenuous.

A transactional chatbot that adds value to a business would be one that, for example, would be able to add money to your travel card. Having a chatbot able to execute the task deflects this common transaction away from the contact centre. This is of value to a business as it allows them to free up agents to deal with more complicated customer issues that require a live agent.

Companies should also feel confident to challenge their conversational AI provider to substantiate what they are saying.  We hear and read a lot about responsible AI, but you need to insist that your provider can show you exactly how they going to deliver this, the measures they have embedded and processes they are following. Responsible AI starts with responsible humans.

Similarly, companies should be confident that their provider is fully versed in data protection, private and sovereignty, and understands the language model infrastructure and how it uses data. Organisations must ask what is happening to the data, and if queries put to ChatGPT are being processed on overseas infrastructure how does this affect data sovereignty or other data protection issues.

The pace at which LLMs are advancing means that the best conversational AI provider is one that is extremely agile in terms of product development. Recently we added the capability to deploy language models locally, both on-premise and in a private cloud, to alleviate the concerns of some clients about sending data to globally deployed models.

Our approach to product development has always been in collaboration with our customers and partners. This ensures that our innovation is driven by solving real business challenges and enables us to deliver real business value.

Unfortunately, there are a lot of unsubstantiated claims about ChatGPT, LLMs and AI but with the right provider, conversational AI is a great tool for businesses looking to deliver better customer and employee experiences.

The excitement about ChatGPT is causing human hallucinations, so ask those questions of your conversational AI provider.