A Detailed Comparison of Conversational AI Platforms and ChatGPT
By Porter L. Versfelt III
The Chat GPT Effect: How Large Language Models are Revolutionizing Conversational AI
Based on an interview with Nick Orlando and Girish Ahnkari
With Open AI’s ChatGPT-3 launch in late 2022, many AI experts are debating the potential uses of the newest artificial intelligence on the scene. It became a trending topic just two weeks after launch, gaining over a million users in that time. Some call what has transpired an AI arms race. It has been called “the Chat GPT effect”.
“There’s a new arms race happening in the world of conversational AI platforms and technology and it revolves around who can implement the most effective use cases and features using Large Language Models.”
The analyst firm Juniper Research writes that ChatGPT and other chatbot messaging apps will come to play an important role in customer interactions in future years. It predicts that high-value online retailers will account for more than 50% of global chatbot spending by 2026.
Nick Orlando, Director of Product Marketing with Kore.ai says, “It’s important to remember that ChatGPT is an impressive research project that has yet to be productized for use in a consumer setting. The language model has been trained on a vast amount of historical, public data from the internet up to 2021, so there are limitations that must be taken into consideration when using it in 2023 and beyond.”
The Battle For The Best Large Language Model
According to Girish Ahnkari, SVP Engineering at Kore.ai, large Language Models (LLMs) are artificial intelligence tools that can read, summarize and translate texts that predict future words in a sentence. “This allows the tech to generate sentences eerily similar to how humans write and talk.”, says Ahnkari.
In late February 2023 Meta Platforms Inc. released to researchers a new large language model (LLM) which is the centerpiece of a new artificial intelligence system, BlenderBot 3. From the get-go it had problems such as offensive remarks and spewing out conspiracy theories.
Nick Orlando says artificial intelligence is more than one technology or breakthrough and involves a collaboration of many different approaches. “There is no doubt that ChatGPT is a project that is giving the public a taste of how useful functional AI will be in their lives. Technology is now at the point where you can build robust “human-like” conversations and actually deliver on the promise of providing great experiences. Which is the beauty of conversational artificial intelligence.”
But ChatGPT cannot deliver that value according to experts in the conversational AI field. Its Large Language Model is limited in what it can do. LLMs are great if you want to search pop culture references or poetry, but developers such as Girish Ahnkari, SVP Engineering at Kore.ai says, “Customers expect swift and effective problem-solving — and that’s why AI-driven virtual assistants stand out as truly intelligent customer service solutions.”
Last December, OpenAI co-creator Sam Altman wrote on Twitter that, “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness. it’s a mistake to be relying on it for anything important right now. It’s a preview of progress; we have lots of work to do on robustness and truthfulness.”
Copyright infringement is one major issue plaguing ChatGPT because its output is based on texts generated by humans. Ethics is another concern. Is it ethical to use ChatGPT in therapeutic counseling or during customer service interactions?
Three Advantages of AI-Powered Intelligent Virtual Assistants (IVA’s) Vs. Basic Chatbots
AI-powered Intelligent Virtual Assistants can have a near-human conversation with you, ChatGPT cannot. Conversational AI offers more human-like voice or text chat conversations in real time. In addition an IVA offers the customer or the company call-center agent more contextual customer experiences. In other words, when someone says or writes “I want to check my account”, the virtual assistant knows that the customer meant ‘check my account balance’.
Conversational AI can be used to create voice and chat-based solutions for everything from customer service, IT support, marketing and digital commerce. There are companies which offer solutions focused on specific enterprises such as healthcare, banking, and on-line retail, such as the Kore.ai XO Platform.
A company can increase employee productivity with Intelligent Virtual Assistants (IVA’s) that talk to its customers. They start by asking the customer questions such as, “What can I help you with today?”, or “Would you like to talk about your overdue credit card payment?” The IVA can drill down and get to the customer’s core needs by using its contextual ability along with a series of questions. It recognizes that there are different ways of saying the same thing.
What Are The Limitations of ChatGPT?
There are some similarities between ChatGPT and conversational AI platforms such as the XO Platform, but they are not interchangeable and are designed to be used for different things. According to Ahnkari, “Conversational AI platforms construct and deploy chatbots and virtual assistants that allow users to interact with them through natural language conversations in order to provide a solution. The limitations for making it a robust, intelligent virtual assistant are substantial.”
“These platforms typically include a range of tools and features that are necessary for any virtual assistant to function effectively. In contrast, ChatGPT is a specific machine learning model that is designed to generate human-like text based on a given prompt or context”, Ahnkari says.
In contrast, OpenAI’s web page admits certain limitation to its ChatGPT:
“May occasionally generate incorrect information.”
“May occasionally produce harmful instructions or biased content.”
“Limited knowledge of world events after 2021.”
National Technology Evangelist Patrick Taylor of AIM Consulting observes that, “In using ChatGPT, we found that it is excellent at open-ended questions like “where is a good place to take my family on vacation?” But when presenting a factual query such as where to go to watch a movie, “it gave answers that were incredibly wrong.”
What Are the Key Limitations of Using ChatGPT as an Intelligent Virtual Assistant?
Lacks Integrations for Transactional Tasks
ChatGPT responds well to customers with its pre-trained knowledge base (LLM), but customers often need more complex transactions which require interacting with a company’s back-end systems.
Does not provide factually correct information
ChatGPT does not have knowledge of the real world and bases its answers on a large dataset that may be outdated, incorrect or just inappropriate. OpenAI admits that its ChatGPT “sometimes writes plausible-sounding but incorrect or nonsensical answers.”
Can’t solve enterprise specific problems
ChatGPT is not able to handle organization-specific queries or FAQ’s on its own. A chatbot or virtual assistant powered by conversational AI is needed to do that. Linking them to a company’s systems and processes allows appropriate quotations and replies for customers.
Not an option for On Premise deployments
Deploying ChatGPT on-premises at your organization isn’t possible because of the size and complexity of the LLM. What are needed are a powerful server or even a group of servers with high-speed network connectivity, and a technician to oversee it all.
Data privacy and security is a concern
LLMs and other NLP models may process and analyze personal data, such as names, addresses, and other identifying information, as part of their operations. Financial or medical information are also sensitive.
Among the privacy concerns are:
⦿ Collection and use of personal data
⦿ Security and confidentiality
⦿ Data storage and retention
⦿ Data sharing and access
Is ChatGPT For Your Organization?
As you have seen here there are many considerations to using an LLM platform such as ChatGPT. Conversational AI platforms such as the KOre.ai XO platform are able to maximize its potential in various areas and expedite bot development. Virtual assistants can recognize the correct intent by understanding the language and semantic meaning of an utterance using its pre-trained LLM. Better test data can be generated, and entity identification can speed up development.
“With its limited capabilities, OpenAI’s ChatGPT may help you in multiple day-to-day activities but does it make a solid replacement for conversational AI technology? The simple answer is a big NO”, says Ahnkari. “While ChatGPT can be used to generate responses that might be used by a chatbot or virtual assistant, it does not have enough features to be able to replace a conversational AI platform on its own”, he says.
You can build your own Intelligent Virtual Assistant using the Kore.ai platform. Try it for yourself.
