The advancements of HR chatbots and their functionality in the gen AI era

Picture yourself in a scenario where a project deadline was misunderstood, and you're unsure  how it might unfold in a broader team setting. A chatbot steps in and says, “Let’s do a role play. I’ll take on the role of a colleague who's frustrated about the missed deadline. You begin  the conversation, and I’ll provide feedback on how you respond.” 

Well, this showcases how Gen AI HR chatbots go beyond traditional capabilities, making HR processes smarter, faster, and more intuitive. 

To begin with, HR chatbots are virtual assistants that are intended to receive inputs in the form  of queries from employees and provide relevant output through dialogue. One of the core  objectives of a HR chatbot is to automate and streamline repetitive HR tasks. 

HR chatbots are utilized across a wide range of functions, including recruitment—where they  assist with gathering information and pre-screening candidates, onboarding - guiding new  hires and managing documentation, and employee engagement, through feedback collection  and event promotion. They also play a role in health and wellbeing by tracking employee  wellness, as well as in training and development by recommending relevant learning  resources—and the list continues to grow. 

Although chatbots have brought numerous advantages to the HR field such as faster response  times, 24/7 accessibility for employees, enhanced productivity, reduced human error, and  personalized interactions, it's equally important to explore their evolution and understand the  current landscape. 

As per a report from IBM, AI-powered conversational agents can address up to 80% of  commonly asked Tier 1 support questions. 

Evolution of HR Chatbots: 

Rule based HR Chatbots were developed that analyses users’ text input based on keywords  and give appropriate responses through pre-defined rules or decision flows. They are not  powered by Artificial Intelligence and progress through a decision tree. While they have their  own advantages, there are a number of limitations such as lack of contextualization, limited  learning and lack of dynamic adaptation due to limited use cases thereby showing incapability  towards complex queries. 

With the evolution of AI based chatbots, usage of advanced technologies of Natural language  Processing and Machine Learning offer unique responses to employees’ queries. The process  starts by building a knowledge base, after which the bot analyses the user's intent and tries to  offer the most accurate response. AI chatbots must undergo a training phase during which a  programmer instructs them on how to interpret the context of a person's language. It is  because of this, AI assistants are able to answer complex queries mimicking a human  conversation. 

The entrance of Generative AI (Gen AI) brings in a revolutionary shift to the world of chatbots. Gen AI doesn’t need to be trained in the same way as traditional AI. Rather than simply  regurgitating information, it can draw on existing knowledge bases, including documents,  websites and public resources like Google Search. Gen AI goes beyond mere repetition by  interpreting information, understanding context, and producing responses that resemble  human communication.

Gen AI Chatbots 

Generative AI support systems efficiently utilize extensive knowledge sources to offer  dynamic, real-time solutions for intricate user queries. 

How do they work? 

Knowledge Base: The bot is programmed to go through reference documents such as PDFs  or any other sources that have been uploaded into the system. It draws pertinent information  based on the query it collects.  

Machine Learning: Machine learning is a key component of AI chatbots, enabling them to understand and  respond to human language, automate tasks, and learn from data 

Natural language Processing (NLP): Natural language processing (NLP) is a type of artificial intelligence (AI) that allows computers  to understand, interpret, and respond to human language. NLP uses machine learning to  process and analyze text and data, and it's based on linguistics, the study of language  meaning. 

AI chatbots make use of Natural Language Processing to analyze inputs, extract key  information and find out the intent behind the query. This allows them to understand user  requests and formulate suitable responses. 

Word/Sentence Embedding: 

Embedding in natural language processing (NLP) is a technique that represents words as  numbers or vectors so that computers can work with them. It maps each word to a vector in a  multi-dimensional space, where the distance between vectors represents the similarity  between words. 

So the reason as to why text is converted to embedding is to perform a similarity search. 

Imagine we have three words: "dog", "cat", and "car". In a word embedding space, these  words might look like this in a simplified 2D plot: 

∙ "dog" = [0.9, 0.8] 

∙ "cat" = [0.8, 0.7] 

∙ "car" = [0.1, 0.2] 

In this case: 

∙ "dog" and "cat" are close to each other because they are both animals. ∙ "car" is far away from both, because it's not related to animals—it’s a different  category entirely (a vehicle). 

This way, word embeddings help computers understand which words are similar based on  their meanings or categories. 

Semantic Search: 

In the context of AI, semantic search refers to the system’s ability to understand the meaning  of the text and process user queries based on the intent and contextual meaning rather than  solely relying on the matching of keywords. Accordingly, the relevant information is extracted 

from the database which in turn allows chatbots to provide clear and contextually appropriate  responses 

Scenario: 

You have a few sentences, and you want to search for information about "food". Sentences: 

1. Sentence 1: "I love to cook pasta with tomatoes and basil." 

2. Sentence 2: "Apples are a great snack, full of vitamins." 

3. Sentence 3: "Eating vegetables is important for maintaining good health." 4. Sentence 4: "I’m thinking of making a sandwich for lunch today." 

Regular Keyword Search vs Semantic Search: 

If you search for "food", Sentence 1, Sentence 3, and Sentence 4 might not show up  because they don’t contain the word "food". Only Sentence 2 may appear, because it  directly mentions "apples", which is a food. 

In semantic search, the system understands that "pasta", "sandwich", and "vegetables" are all types of food. So, even though the word "food" isn't used directly, Sentence 1,  Sentence 3, and Sentence 4 will still show up because they talk about food-related items. 

Large Language Models (LLMs) & Langchain: 

LLM stands for Large Language Model, which is a type of artificial intelligence (AI) program  that uses deep learning to analyze and generate text 

LLMs are trained on large amounts of text data, such as millions of gigabytes of text from the  internet. This training process uses a neural network, which is then used to perform tasks like  generating or translating text. LLMs have the capability to understand and generate text,  interpret human language, summarize, translate, predict, and create content. 

Langchain is an open source framework that assists developers in creating applications that  use large language models (LLMs). It offers a framework for simplifying the development  process, including pre-built tools and components, a standardized interface, and memory  management. Langchain is a good choice for developers who want to prioritize ease of use,  rapid development, and cost-effectiveness. 

Models such as GPT (Generative Pre-trained Transformer) are capable of producing human like text based on the input they are given. Large Language Models (LLMs) empower AI  chatbots to grasp the context of user queries and produce responses that are contextually  relevant. 

In conclusion, while Gen AI is evolving and limitations such as hallucinations (false  or misleading information) are prevalent, it still has a lot to offer and transform the way HR  functions, providing a better employee experience. 

The future of AI chatbots in HR is set to make HR functions more efficient, personalized, and  data-driven. By automating routine tasks, enhancing employee engagement, and providing  valuable insights, AI chatbots will help HR teams focus on more strategic and high-value  initiatives while improving the overall employee experience. As AI technology continues to evolve, chatbots will become even more intuitive, empathetic, and integrated, making them  indispensable tools in modern HR management.