The launch of ChatGPT instituted our understanding of conversational AI, albeit it had existed for some years before OpenAI’s product spearheaded the idea into the mainstream.
But as this year has unfolded, it’s become clear that conversational AI is also a foundation stone for the next evolutionary periods of artificial intelligence to build upon. From Microsoft Copilot to Zoom AI Companion to Google’s Duet AI — as we approach the end of 2023, it is generative AI that now has centre stage.
“Generative AI is the next generation, the next level of AI,” Jon Arnold of J Arnold & Associates, told UC Today at October’s UC Expo 2023 in London. “Up until gen AI, it’s really been about advanced search capabilities where you give the inputs, and it finds stuff for you. Generative is now it has the inputs, and it’s giving stuff back to you; it’s generative. It’s producing all kinds of new ways to automate things, which is really handy.”
Arnold was enthusiastic about Gen AI’s use cases in contact centres and said, “The cool thing about it is the more you train AI tools, the better the generative outputs will be”.
“It’s a little hard to trust right now where it’s headed, and it’s so easy to have what they call hallucinations, which basically could be gibberish,” Arnold clarified. “It thinks it’s giving you something that makes sense, and maybe with the syntax and grammar it is.”
There has been scepticism of Gen AI, however, rooted in its misuse of data. ChatGPT extracts user prompts to develop and improve its model unless users deliberately opt out. This has prompted worry that employees might accidentally include proprietary or confidential data or information in their prompts, which ChatGPT leverages to answer future queries.
Data leaks have been common. In June, OpenAI was subject to a class action lawsuit filed in California federal court, saying it extracted “massive amounts of personal data from the internet”. The allegation claimed that OpenAI stole and misappropriated millions of peoples’ data from across the internet to hone its AI models. In June, Microsoft established legal protections in the form of AI customer commitments to assuage concerns about data leaks, such is the seriousness of the issue.
At this stage of Gen AI’s broad development across each vendor and platform, does Arnold believe the use of business data can be trusted to these services?
“The short answer is ‘No’, but we’re going to have to learn how because the possibilities when you get it right are just too good,” Arnold said. “AI has scale and speed that we cannot match any other way. The key to this, and why generative AI is interesting, is this whole shift in digital transformation, meaning that the more digital inputs that we have, the more data points AI has to learn and get better with.”
“So it may not be great today, but the volume of data it’s capturing now is growing exponentially, mainly because the cloud is a great cost-effective way to store data so you can keep adding, adding, adding, and it will get smarter and smarter.”
If generative AI is the major trend as 2023 approaches its end, what might 2024 have in store?
“Even though the year’s almost over, I still have eight more events to attend, so my answer might change around mid-December,” Arnold caveated, “but one idea I want everyone to consider, the acronym of the week or month, is LLM (large language model). Just like we had conversational AI, now we have generative AI. What this is leading to is the use of AI to consolidate information and learning across an organisation.”
“The term large language model is at the root of this,” Arnold continued. “The more data you collect within your organisation about your market, the more precise you can be in understanding things. Companies are starting to develop proprietary large language models, and it’s a bit of a walled garden thing, but it’s their way of almost using data from AI as a competitive advantage.”
Arnold highlighted that if an organisation has language models specific to its customer base, organisation and culture, “its ability to communicate becomes very powerful, and it can be a way of doing it better than your competitors if they aren’t as advanced in their LLMs as you are, so it can have tremendous value and a way that you harness the data that’s been captured”.
Are We Undergoing a CPaaS Renaissance?
Arnold also discussed the revival of CPaaS in 2023, with several stalls and talks at this year’s UC Expo covering the subject.
CPaaS as a concept has been around for years, but it has often been quite an ambiguous idea, as Arnold noted: “What is CPaaS? I don’t think you can still get a handle on it. It’s a little like what UCaaS used to be because UCaaS is a vendor term. You don’t really know what it means, and CPaaS is even harder.”
However, as he highlighted, Arnold was excited by the underlying capabilities of CPaaS, many of which are driven by AI.
“They are really powerful because the idea is this term of ‘Programmable communications’,” Arnold explained. “You can customise applications on the fly with CPaaS; it integrates with all your platforms, and that’s the key that makes it really good because now you can have the contact centre because we have what you call proactive notifications.”
“Say your store hours change because of a weather issue or something, and you can push those notifications almost in real time when you have the ability to code on the fly. There is this term, ‘No code, low code’, so that means it’s very much a user-generated thing. You don’t have to have expensive developers on staff, which opens up things in a big way.”
UC Today’s video interview with Jon Arnold will be released in the near future.
