Applying ChatGPT in customer contact – the intermediate position

by Erik Bouwer

Applying ChatGPT in customer contact – the intermediate position

by Erik Bouwer

by Erik Bouwer

How can you apply ChatGPT in customer contact? Several CX providers have already developed services. Ziptone draws up the (interim) balance.

This article is regularly updated and is part of it File Chatbots

 

It took TikTok nine months, Instagram took 2,5 years, but ChatGPT reached a volume of 2023 million active users in five days (in the month of January 100). This is according to Reuters the fastest growing application of all time. Speed ​​is also ChatGPT's main achievement: the tool can generate near-real-time smooth texts based on prompts. In addition to that speed, another important feature is that the tool does not have to be trained endlessly to do its job.

Remains about reliability. If the output of ChatGPT (and related solutions) is guaranteed, a range of solutions is expected to be within reach. The race between Google ('Bard') and Microsoft (via the billion-dollar investments in OpenAI) will undoubtedly cause further acceleration.

Paid variant

Parent company OpenAI will in any case come with a paid variant of ChatGPT, under the name ChatGPT Plus. The monthly costs are about $ 24 and that version will also be available in the Netherlands and Belgium. The paid version offers extra functionality, a faster response time and faster access to the chatbot. In addition, pay-as-you-go purchase models are available with different characteristics, where payment is made on the basis of tokens. Those tokens are again based on the number of words that are generated or processed.

Which ChatGPT applications are offered?

So far on the developer(s) side. The necessary work is also being done on the side of service providers. Although all still modest in design, there are already quite a few software vendors who integrate ChatGPT functionality into their own offerings. We list the possibilities for applying ChatGPT in customer contact for you.

Integration capabilities in Azure ecosystem – Microsoft – in addition to being an investor in OpenAI, also owner of speech recognition platform Nuance – will build ChatGPT functionality into Office applications and into the Bing search engine. In doing so, Microsoft serves several purposes: in the first place, broadening both adoption and daily use. And further development of the tool in the Azure environment. After AWS (33%), Azure is the most important public cloud with a market share of 21% (with a total of four million servers and 40 million hard disks). Within the Azure Marketplace nearly 200 AI-powered chatbot solutions alone are available. Companies primarily choose AWS or Aure or both in their cloud strategy; the platform is also the place where applications are integrated. In short, the Azure ecosystem is something to keep an eye on.

Baidu (the Chinese counterpart of Google) is also launching a ChatGPT-like solution this year. Baidu states that this chatbot is based on a language model that is significantly larger than GPT-3. Baidu's chatbot will be named 'ERNIE Bot'. This refers to the model used: 'Enhanced Representation through Knowledge Integration'. ERNIE is expected to be presented to the public in March.

Agent Assist – Observe.AI, provider of a conversation intelligence platform, comes with agent-assist applications for the contact center. The Observe.AI solution supports contact center employees in customer interactions. The real-time AI of Observe.AI listens from start to finish and gives advice on the fly. This contributes to up- and cross-selling, higher customer satisfaction and a higher degree of compliance, according to the company.

Chatbot optimization – NICE has announced that it will integrate its CXone Expert solution with ChatGPT. The integration will not only lead to more efficient and accurate chatbot responses, but also to a more human interaction. CXone Expert is NICE's cloud-based knowledge management solution, which leverages NICE Enlighten AI models, designed specifically for CX use cases. The company also leverages organization-specific data to help customers create custom conversational AI models.

Translate real time – The Dutch customer service software company Belco will use the AI ​​language model ChatGPT to translate conversations within their digital customer service environment into other languages. Customer service employees regularly receive questions via various (online) channels in a language other than Dutch. With the new link between Belco and ChatGPT, these questions can be quickly translated into Dutch. Answers to questions can also be easily translated by the AI ​​language model.

Analytics in your own language – Also CX analytics platform Glassbox has integrated ChatGPT so that users can ask analytics questions in their native language.

Generate summaries – Talkdesk recently launched call summarization, integrating GPT services into its NLP systems. This feature automates call processing for agents, leveraging the generative AI solution's capabilities to summarize and classify call transcripts. A variant of this is used indoors ContactCare, which optimizes call logs.

Adjust Tone or Voice – Intercom has integrated ChatGPT into its customer service platform. This allows Intercom users to use the AI ​​to generate answer suggestions for customers based on a few keywords. Another application that Intercom offers is to adjust the tone of an answer with a few clicks, for example for specific target groups. Intercom will also use the application to transfer customer cases between employees by generating a summary of the conversations held.

Interpreting Conversations – Five9 has released two new products that leverage OpenAI's ChatGPT-3: AI Insights and AI Summaries. AI Insights combines ChatGPT-3 with real-time transcription to automatically interpret and categorize customer conversations. By grouping contacts – based on intent or other characteristics – contact centers can quickly identify opportunities for automation and process improvement. AI Summaries provides automatically generated summaries of conversations, after which they can be recorded in the CRM environment. In addition to these options, Five9 also offers modules for speech analytics, workflow automation, and IVA solutions.

Nuance also embraces ChatGPT – In the Nuance use case, ChatGPT can take over a conversation when the consumer asks a question for which the bot is not trained.

Salesforce Introduces Einstein GPT – Einstein GPT can, among other things, generate personalized emails and create knowledge articles based on conversation reports from previous cases.

Much more possible

Applications can also be expected in other areas in the near future. Think of the automated generation of spoken or written solutions; generating user manuals, tutorials and how-tos to guide customers through processes or sub-steps; and emotion detection in spoken language or the automated response to emotions in written language.

Downsides of ChatGPT

Disadvantages that have been known for some time are the possible copyright infringements, the bias and limited use of resources, and the lack of transparency: although OpenAI says it adheres to GDPR rules, it is not clear where the input that users give ends.

Security experts also point out security risks. ChatGPT has been developed in such a way that the tool actively blocks requests for generating content that targets criminal activity. ChatGPT will not respond to requests to write code to steal data from a hacked device or create a phishing message. But hackers have already figured out ways to get around those restrictions. They already offer their illegal services on underground crime forums. Among other things, the hackers use the API of OpenAI's DaVinci model, which at the beginning of February was not yet protected against criminal intent.

How sustainable is real-time AI?

And finally, it is still unclear what the energy consumption of OpenAI applications looks like. In addition to searches and general internet usage (good for about almost 4 percent of global emissions), real-time processes are huge energy consumers. According to Wired the switch to AI-based searches such as Google and Microsoft envision could lead to a five-fold increase in computing power. (zip tone)

This article is being supplemented.

Follow by Email
Whatsapp
LinkedIn
Share

Also interesting

Featured, Knowledge base, Technology
Top