How OpenAI’s ChatGPT and LLMs like GPT-4 are transforming customer support chatbots

Natalie Smithson
AI enthusiast | Tea addict | Focused on using AI assistants to win the working week
AI assistant customer service

If like millions of other people you’ve enjoyed playing around with ChatGPT and learning about GPT-4 (the large language model, or LLM, that supports it), you’ll love the next two minutes on how this revolutionary technology is set to transform customer support.

What are LLMs like GPT and LaMDA?

Recently, OpenAI’s ChatGPT has created huge awareness around text-generating AI. It uses a type of machine learning called a large language model (LLM) ― in this case, one called GPT-4, which can handle a wide range of natural language processing (NLP) use cases. The speed and creativity with which ChatGPT can answer questions, write stories, and create sophisticated essays and documents has taken the world by storm. Google has launched Bard, which uses LaMDA as its LLM, and Microsoft are working with ChatGPT to evolve their search engine, Bing. The big question in customer support is, how transformational will this be for the next generation of chatbots?

Launch a chatbot in 15 minutes and go even faster with an LLM

AI assistants like ours are already in the process of displacing traditional chatbots because of their speed, low cost and accuracy. A ready-to-launch AI assistant can be live in 15 minutes, to start cutting contact volumes, tackle staff shortages and offer a 24-hour service. Now, we’re using LLMs like GPT-4 to automate training alongside our conversation review service, which keeps everything on track.

Using LLMs to help train your AI assistant puts rocket fuel in the process

Our average success rate for the advanced AI assistants we create is 91%. That means your AI assistant has recognised the request from the customer, understood what they need, and given a helpful response without any input from your customer contact team.

Once you go live, our expert team of language experts, psychologists and writers review conversations customers are having with your AI assistant ― all PII (personal information) is removed. They’re the ‘human in the loop’ helping you achieve a 99% success rate (allowing 1% for questions it hasn’t learnt the answer to yet, but soon will).

Now, we’re using GPT-4 to help you hit a 99% success rate in record time. You don’t need to lift a finger or know anything about AI, LLMs, GPT-3, 3.5, or GPT-4 to benefit.

You can train your AI assistant without needing to brainstorm

Whereas a traditional chatbot needs a lot of upfront training and input from your team, an advanced AI assistant only needs a handful of phrases to launch from. We call these flows.

When you add a flow, you’ll want to give your AI assistant a few different examples of how a question might be phrased by different people. “I want to buy a t-shirt,” “what t-shirts are available,” and “have you got any T-shirts on sale” are all effectively the same question. If you can give your AI assistant a few different examples of phrasing to learn from, it’ll soon pick up the link.

Our current advice is to use FAQs or existing customer contact for phrasing ideas. You only need a few minutes to complete this step, but sometimes people get stuck coming up with ideas or worry about writing the ‘wrong’ thing. Now, you can use GPT-4 within our platform to generate alternate phrasing ideas instantly. An LLM like GPT-4 has absorbed so much information from so many different places, it’ll never be short of suggestions to keep you moving through the process. Just press a button and we’ll pull in its suggestions for you to accept, edit or reject ― nothing goes into a flow without your say-so. 

Take advantage of all the latest LLM releases from the safety of our platform

Watching people experiment with ChatGPT in particular, you might be nervous about accuracy. We are too. That’s why trusted and experienced AI providers won’t be suggesting LLM content feeds directly into your live customer conversations ― yet. There’s potential for this in the future, but it’s a long way off.

We’ve put eight years’ worth of AI knowledge and 21+ years of data engineering into producing AI Studio where you can easily create and manage your own AI assistant. Thanks to our human in the loop, whether you use your own knowledge or lean on LLM for phrasing suggestions, your AI assistant will operate at a safe level of accuracy from the minute it starts talking to your customers because our human in the loop team is always there in the background, guiding you.

GPT-4 is the latest release from OpenAI (with others coming thick and fast), which will make it possible for us in future to create your AI assistant almost instantly. Collating content you’ve already produced for your business (like web copy, social channels, emails and FAQs) and feeding it in a flash to your AI assistant, making it an instant expert on your brand, business and customer needs. On our platform, you’re still the one to check everything over and push content live to your AI assistant ― you stay in full control.

It’ll be exciting to see the wider business community learn more about applying the immense power of LLMs. We’ll implement new AI releases so you can safely benefit from them without the risk that comes with experimentation, so you can focus all your attention on giving the most sophisticated level of customer support and transform your customer experience.