Learning Library

← Back to Library

Rules vs AI vs Generative Chatbots

Key Points

  • Rules‑based chatbots follow rigid, keyword‑driven flows that often fail when customers deviate from pre‑programmed scripts, leading to misunderstandings and lost sales.
  • AI‑powered chatbots with natural language understanding can interpret varied phrasing, personalize interactions, and seamlessly integrate offers and customer data for smoother transactions.
  • When a customer’s issue escalates—such as a complaint about a cold pizza—the AI chatbot can recognize dissatisfaction and promptly transfer the conversation to a live agent, preserving the relationship.
  • The optimal chatbot type for a business depends on the complexity of tasks, required personalization, and the need for real‑time human handoff, making the choice “it depends.”

Full Transcript

# Rules vs AI vs Generative Chatbots **Source:** [https://www.youtube.com/watch?v=93l-2Usc08U](https://www.youtube.com/watch?v=93l-2Usc08U) **Duration:** 00:05:16 ## Summary - Rules‑based chatbots follow rigid, keyword‑driven flows that often fail when customers deviate from pre‑programmed scripts, leading to misunderstandings and lost sales. - AI‑powered chatbots with natural language understanding can interpret varied phrasing, personalize interactions, and seamlessly integrate offers and customer data for smoother transactions. - When a customer’s issue escalates—such as a complaint about a cold pizza—the AI chatbot can recognize dissatisfaction and promptly transfer the conversation to a live agent, preserving the relationship. - The optimal chatbot type for a business depends on the complexity of tasks, required personalization, and the need for real‑time human handoff, making the choice “it depends.” ## Sections - [00:00:00](https://www.youtube.com/watch?v=93l-2Usc08U&t=0s) **Untitled Section** - - [00:03:21](https://www.youtube.com/watch?v=93l-2Usc08U&t=201s) **Generative AI Elevates Customer Experience** - The transcript demonstrates a generative‑AI‑powered chatbot personalizing a greeting, recommending complementary items, verifying allergen safety, and showing source‑backed answers, illustrating how AI can make interactions more fluent, empathetic, and trustworthy. ## Full Transcript
0:00Have you ever interacted with a chatbot that can't  seem to figure out what you need and responds with 0:05"I'm sorry I don't understand. Please try again."  Yeah, we've all been there with a rules-based 0:11chatbot, when a simple task of ordering something  turns into a headache. Are rules-based chatbots 0:17obsolete? No! But let's look at three different  kinds of chatbots with various degrees of 0:22sophistication: the rules-based, AI-powered, and  generative AI. Which is the right kind for your 0:30company? The answer is "it depends". Let's look at our first example, 0:34the limited rules-based chatbot, which walks  the customer through a rigid, predetermined 0:40flow and operates on basic keyword detection. [chatbot] Thanks for choosing Taste of Italy. 0:45Our pizzas are yummy! How many  pizzas would you like to order? 0:48[Bella] Two. [chatbot] What size would you like 0:50them to be? Small, medium, or large. [Bella] Large. 0:55As you can see, there's not much AI  involved in this interaction. Of course, 0:59placing an order is not always so straightforward.  Because it's impossible for the conversation 1:05designer to predict and pre-program the chatbot  for all the ways a person can place an order, 1:09this kind of chatbots often get stuck. [Bella] My pizza is cold! I want to file a 1:16complaint and get my money back! [chatbot] Sorry, I don't understand, 1:19try rephrasing your message. Also, if the customer has a 1:24complaint about the order, the chatbot  could act as a gatekeeper. This is a sure 1:28way to send customers to competition.  Let's start the conversation again, 1:33this time with an AI-powered chatbot! [Bella] I want to order two large 1:39pepperoni and cheese pizzas. [chatbot] Great choice, 1:41Bella! Would you like any extra toppings? [Bella] No, just the regular, thanks! Add 1:46one apple juice and one beer to the order. With AI and natural language understanding [NLU] 1:51capabilities, the chatbot can understand the  customer's request, the first time—no matter 1:57how it's phrased. With deep learning, the longer  an AI chatbot has been in operation, the better 2:03it can understand the intention behind questions  and provide more detailed, accurate responses. 2:09Let's continue: [chatbot] We have a new offer – order 2:121 pizza and 1 dessert, and get a drink for free.  Would you like to take advantage of this? 2:16[Bella] Of course, who doesn't like dessert? [chatbot] Where would you like this order to 2:20be delivered? [Bella] Home 2:21[chatbot] Confirming 2 promotional packages. 2:24Delivery in 45 minutes. Enjoy dinner, Bella! 2:27Deeply integrated with the business systems, the  chatbot knows the customer's name and address, 2:32presents the latest offers  that might be of interest, 2:35and creates a smooth ordering process. [Bella] My pizza is cold! I want to file 2:41a complaint and get my money back! [chatbot] I am sorry to hear that. I 2:44am transferring you to a live  agent who can help you now. 2:47When the customer is unhappy and wants to speak  to a human, the transfer should happen seamlessly. 2:54After being transferred, the live support agent  will get the chatbot conversation history and can 2:59start the call informed. But what would this  look like with a generative AI chatbot? Would 3:06the customer want to read a  chatbot-generated poem about 3:09pizza when placing an order? [Bella] I want to order 3:12two large pepperoni and cheese pizzas. [chatbot] In this vibrant city's culinary scene, 3:17lies a pizza divine, a flavor supreme.  Would you like to order a pizza today? 3:21Hmm, that's certainly added personality, but  probably not all that helpful in this case. 3:28Ok, let's start over. How can generative  AI elevate the customer's experience? 3:34[Bella] Today I'm celebrating my  college graduation! I want to order 3:38two large pepperoni and cheese pizzas. [chatbot] Nice to see you again Bella, 3:42and congratulations! No party meal is complete  without something sweet! We have the most 3:47delicious tiramisu cake and a new offer – order 1  pizza and 1 dessert, and get a drink for free. Would 3:53you like to take advantage of this? [Bella] You bet! 3:57The next generation of generative AI chatbots can  get even smarter and more fluent in understanding 4:03common language, adapting to the customer's style  of conversation, and using empathy when answering. 4:09You can optimize the chatbot response even further  using generative AI capabilities for search. 4:15Let's see how: [Bella] I have 4:18an allergy to peanuts. Can you confirm that  this dessert does not contain any peanuts? 4:22[chatbot] This tiramisu does not contain  any ingredients with peanuts. I searched 4:27my knowledge base. You can expand to see  all the ingredients and what they contain. 4:31Chatbots with conversational search capabilities  can give you answers to the questions you're 4:36looking for, while showing you the source if  you want to validate the chatbot's response. 4:41So, do chatbots need AI? Well, yes, it does make  chatbots smarter. They can better understand 4:48intent and respond more accurately to questions  by learning from its previous interactions, 4:53improving customer experiences. When thinking  about generative AI's impact on chatbots, 4:58think about how you can take advantage of  creative, conversational responses and when this 5:04technology makes the most sense for your business  objectives and the needs of your customers. 5:09Thanks again for watching. 5:10Head to watsonx Assistant to learn more about how to leverage AI chatbots in your business.