Textual Intelligence

Meenakshi Malik, PhD is an assistant professor at BML Munjal University, India, with over 12 years of experience. She earned her computer science and engineering doctorate at Maharshi Dayanand University, Rohtak, in December 2023 and was honored by the Vice President for her exceptional PhD research. Her research interests include artificial intelligence, machine learning, deep learning, and big data. Preeti Sharma, PhD is a faculty member at Chitkara University, Punjab, India. She is the author or co-author of over 12 publications in national and international journals and conferences. Dr. Sharma’s research interests include extensive work in blockchain and its diverse applications, as well as artificial intelligence and machine learning. Susheela Hooda, PhD is an associate professor in the Department of Computer Science and Engineering, Chitkara University, Punjab, India. She has published over 30 technical research papers in national and international journals and conferences. Her research interests include software engineering, aspect-oriented software development, software testing, cloud computing, artificial intelligence, and machine learning.

• Preface xixPart 1: Introduction 11 Introduction: Overview of Generative AI and Multifaceted Applications, Significance, and Potential of LLMs 3K. Mukheja, S. Mittal, C. Monga and S. Annam1.1 Introduction to Generative AI and LLM 41.2 Applications of Generative AI 61.2.1 Medical 61.2.2 Education 71.2.3 Finance 71.3 Detail Case Study—Rise of Chatbots 91.3.1 Empowering Chatbots with Large Language Models 101.3.2 Chatbots in Medical and Healthcare Education 101.3.3 Chatbots in Finance 111.3.4 Chatbots in Tourism 111.4 Examples 121.5 Comparative Analysis of Generative AI Techniques 141.6 Future Scope and Potential 161.7 Conclusion 17References 172 A Comprehensive Study of Large Language Models 21Pawan Kumar, Anu Chaudhary, Shashank Sahu, Mradul Kumar Jain and Updesh Kumar Jaiswal2.1 Introduction 222.2 Background 242.2.1 Tokenization 242.2.2 Positions Encoding 242.2.3 Attention in LLM 252.2.4 Activation Function 262.2.5 Data Preprocessing 262.2.6 Architecture Model 272.2.7 Pre-Training 282.2.8 Fine-Tuning 292.3 Large Language Models (LLMs) 312.3.1 BERT (Bidirectional Encoder Representations Transformer) 312.3.1.1 BERT Architecture 312.3.1.2 Working of BERT Model 322.3.1.3 Fine-Tuning in BERT 332.3.1.4 BERT Applications 342.3.1.5 Advantages of the BERT Language Model 352.3.1.6 Disadvantages of the BERT Language Model 352.3.2 ChatGPT (Chat Generative Pre-Trained Transformer) 362.3.2.1 ChatGPT Architecture 362.3.2.2 Tokenization 382.3.2.3 Embeddings in ChatGPT 392.3.2.4 Pre-Training 392.3.2.5 Fine-Tuning 392.4 Challenges and Future Directions 402.5 Conclusion 40References 41Part 2: Generative AI Project Lifecycle 453 A Deep Learning Methodology with Transformers LLM to Calculate the Global Temperature Difference in Recent Years 47Ana Carolina Borges Monteiro, Reinaldo Padilha França and Rodrigo Bonacin3.1 Introduction 483.2 Overview of Literature IoT 503.3 Overview of Literature AI 533.4 Methodology 563.5 Results 573.6 Discussion 613.7 Conclusions 63References 644 Navigating the Generative AI Project Ecosystem with a Focus on Addressing Data Architecture Complexities and Strategic Model Selection for Optimal Outcomes 67Mohammad Shabaz, Shanky Goyal, Ismail Keshta, Mukesh Soni and Vijay Kumar4.1 Introduction 684.2 Literature Review 694.3 Proposed Method 724.4 Result 834.5 Conclusion 88References 895 Generative AI Project Life Cycle—Use Case Planning and Scope Definition 93Jyoti Rani, Pawan Kumar and Nidhi Sharma5.1 What is Generative AI? 945.2 What is Artificial Intelligence? 955.2.1 Introduction to Generative Life Cycle 955.3 Generative AI on AWS 985.4 Why Generative AI on AWS? 995.5 How is Generative AI Operational? 1015.6 Multiplicative Artificial Intelligence Interfaces 1025.7 ChatGPT 1025.7.1 How Does ChatGPT Work? 1025.7.2 In What Ways is ChatGPT Being Helpful for Users? 1035.8 What Advantages Does ChatGPT Offer? 1045.8.1 What are ChatGPT’s limitations? To What Extent is it Accurate? 1055.9 Dall-e 1065.9.1 How DALL-E Works 1065.9.2 How Do You Use DALL-E? 1075.9.3 How is DALL-E Taught? 1085.9.4 The Prospects of ChatGPT and Generative AI 1095.9.5 Fields that Utilize DALL-E 1105.9.6 Advantages of Using DALL-E to Create Images 1115.9.7 DALL-E’s Effect on Image Production 1125.9.8 Constraints with DALL-E 1125.9.9 Examples of DALL-E’s Use in the Real World 1135.9.10 What DALL-E’s Challenges Are 1135.10 Bard 1145.10.1 What is LaMDA? 1145.10.2 How is Google Bard AI Used? 1155.10.3 Google Bard AI Features 1155.10.4 Examples and Use Cases for Google Bard AI 1155.10.5 AI’s Reach with Google Bard 1165.10.6 Bard AI by Google vs. ChatGPT 1165.10.7 Constraints with Google Bard AI 1175.10.8 Important Uses of Generative AI 1185.10.9 Creation and Manipulation of Images 1185.11 Coding and Software 1195.12 Making of Videos 1195.13 Creating and Condensing Text 1195.14 Interorganizational Cooperation 1205.15 Enhancement of Chatbot’s Performance 1205.16 Business Exploration 1215.17 Conclusion 121References 1226 Generative AI Unleashed: A Multi-Domain Journey of Successful Implementations of Large Language Models 125Nikhil Kumar, Anurag Barthwal, Saurabh Mishra and Abhishek Jain6.1 Introduction 1266.1.1 Background and Motivation 1266.1.1.1 Neural Networks and Deep Learning 1276.1.1.2 Transformers 1276.1.1.3 Pre-Training and Fine-Tuning 1276.1.1.4 Scaling 1276.1.2 Scope and Objectives 1286.2 Literature Review 1286.2.1 Historical Development of Generative Artificial Intelligence 1296.2.2 Evolution of LLMs 1296.2.3 Applications of Generative AI Across Different Domains 1306.2.4 Challenges and Limitations in Implementing LLMs 1316.3 Methodology 1316.3.1 Research and Design 1316.3.2 Methods of Data Collection 1316.3.3 Model Selection and Training Techniques 1326.3.4 Evaluation Measures 1326.3.5 Ethical Considerations 1326.4 LLM-Based Case Studies 1326.4.1 Natural Language Generation in Healthcare 1336.4.1.1 Case Study 1: Patient Diagnosis Support System 1336.4.1.2 Case Study 2: Electronic Health Records Summarization 1346.4.2 Creative Content in Media and Entertainment 1346.4.2.1 Case Study 3: A Scriptwriting Support Tool 1346.4.2.2 Case Study 4: Developing Virtual Characters 1356.4.3 Language Translation and Multilingual Communication 1356.4.3.1 Case Study 5: Multilingual Communication Platform 1356.4.3.2 Case Study 6: Real-Time Interpretation Service 1366.5 Results and Analysis for LLMs 1366.5.1 Performance Evaluation of Implemented Models 1366.5.1.1 Quantitative Metrics 1376.5.1.2 Qualitative Analysis 1396.5.2 Impact Assessment of LLMs Across Different Domains 1396.5.2.1 Impact Assessment of LLMs in Healthcare 1406.5.2.2 Impact Assessment of LLMs in Infotainment 1416.5.2.3 Impact Assessment of LLMs in Language Translation 1426.5.3 User Feedback and Acceptance 1436.5.3.1 A/B Testing: Choice as a Coping Strategy 1446.5.3.2 Surveys: Capturing Broad Feedback 1446.5.3.3 User Interviews: Getting Into the Weeds of UX 1446.5.4 Comparison with Existing Systems 1456.6 Discussion 1456.6.1 Understanding the Successful Implementation of LLMs 1456.6.1.1 Multimodal Generative AI: Unleashing the Power of Many Data Types 1466.6.2 Challenges and Limitations 1476.6.3 Ethical Implications and Responsible AI Practices 1486.6.4 Future Directions and Emerging Trends 1496.6.4.1 LLMs: A Powerful Tool, But One That Demands Careful Consideration for Society 1506.7 Conclusion 151References 152Appendix 155Glossary 1557 Misbehaving AI Models and AI Interaction Issues with Humans 157Nishi Gupta and Shikha Gupta7.1 Introduction 1587.2 Literature Review 1607.3 Misbehaving AI Models 1627.3.1 Causes of Misbehaving AI Models 1627.3.2 Consequences of Misbehaving AI Models 1647.3.3 Mitigation Strategies That Can Be Employed to Address Misbehaving AI Models 1677.4 Human Interaction with AI models 1687.4.1 Human Interaction Issues with AI Models 1687.4.2 Laws Made to Deal with Misbehaving AI Models 1697.4.3 The Importance of Ongoing Research and Development in Addressing Misbehaving AI Models 1717.5 Conclusion 173References 1748 Decoding Potential of ChatGPT: A Comprehensive Exploration of AI Generated Contents and Challenges 177Anju Kaushik and Anil Kaushik8.1 Introduction 1788.2 Chapter Organization 1798.3 ChatGPT Popularity Statistics 1798.4 Implementation and Work Flow of ChatGPT 1808.5 ChatGPT Key Characteristics in Present Scenario 1828.6 Potential Challenges 1868.7 Security Threats in ChatGPT 1878.8 ChatGPT’s Privacy Risks 1898.9 Ethical Concern 1928.10 Computer Ethics Challenges Raised by ChatGPT 1948.11 Limitation of ChatGPT 1958.12 Balance Between Human Knowledge and AI-Supported Innovation 1968.13 Future Challenges 1978.14 Conclusion 197References 1989 Economizing Large Language Model Training and Alignment with Human Values through Cost Effective Architectures and Transfer Learning Techniques 201Mohammed Wasim Bhatt, Rubal Jeet, Mukesh Soni, Haewon Byeon and Vishal Sagar9.1 Introduction 2029.2 Literature Survey 2039.3 Proposed Method 2059.4 Results 2169.5 Discussion 2199.6 Conclusion 219References 220Part 3: In-Context Learning/Prompt Engineering 22310 From Prompts to Performance: Innovations in Context Learning 225Amandeep Sharma, Prince Kumar and Shashank Dhamija10.1 The Art of Prompt Engineering: A Deep Dive 22610.1.1 Core Definitions and Key Concepts of Prompt Engineering 22610.1.1.1 Significance of Prompt Engineering 22610.1.1.2 Fundamental Components of a Prompt 22610.1.1.3 Prompt Engineering’s Technical Aspects 22810.2 Strategies for Crafting Effective Prompts 22910.3 Techniques for Controlling the Model Behavior and Output 24510.4 Best Practices for Prompt Engineering 24610.4.1 Prompt Engineering Principles 24710.4.2 Structured Procedure Behind Prompt Engineering 24710.4.3 Prompt Engineering Use Cases and Applications 248References 250Part 4: LangChain Framework 25311 Introduction to LangChain Framework 255Deepti Goyal and Amita Gautam11.1 Introduction of LangChain Framework 25611.2 Large Language Model (LLM) 25811.3 What Do You Mean by Chains in LangChain Framework 26011.3.1 Various Types of Chains 26011.3.1.1 LLMChain 26111.3.1.2 Router Chain 26111.3.1.3 Sequential Chain 26211.4 Why LangChain Framework is Important 26311.5 Main Components of LangChain Framework 26411.5.1 Large Language Model (LLM) 26411.5.2 Prompt Template 26511.5.2.1 Indexes 26511.5.2.2 Retriever 26511.5.2.3 Parsers for Output 26511.5.2.4 Vector Store 26611.5.2.5 Agents 26611.5.2.6 Memory 26611.5.2.7 Chain 26711.6 Feature of LangChain Framework 26711.6.1 Scalability 26711.6.2 Improved Usability 26711.6.3 Adaptability 26711.6.4 Extension 26711.6.5 External Integrations 26811.6.6 Thriving Community 26811.6.7 Flexibility Across Zones 26811.6.8 Integrations 26811.6.9 Standardized Interfaces 26811.6.10 Prompt Management and Optimization 26811.6.11 Visualization and Experimentation 26811.7 How to Install 26911.7.1 Steps to Develop an Application in LangChain Framework 27011.7.1.1 Describe the Use Case 27011.7.1.2 Develop Functionality 27011.7.1.3 Tailor the Functionality 27011.7.1.4 Optimizing LLMs 27011.7.1.5 Data Purification 27011.7.1.6 Experimenting 27111.7.2 Build a New Application with LangChain Framework 27111.8 Real World Applications with LangChain Framework 27211.8.1 LangSmith 27211.8.2 Chatbots 27211.8.3 Automated Blog Outlines 27211.8.4 Integration with MongoDB Atlas 27211.8.5 Medical Care 27211.8.6 Help with Coding 27311.8.7 Creating Condensed Content 27311.9 Integration of LangChain Framework 27311.10 Creating a Prompt in LangChain Framework 27411.10.1 Types of LangChain Prompts 27511.10.2 Prompt Template 27511.10.3 Few_Shot_Prompt_Template 27611.10.4 Chat_Prompt_Template 27611.11 Future of LangChain Framework with AI Enabled Tools 27811.11.1 ChatGPT and Chatbots 27811.11.2 AI-Powered Text Categorization Tools 27811.11.3 False References 27911.12 Limitation of LangChain Framework 27911.13 Alternative Technologies Apart from LangChain Framework Used in 2024 28011.13.1 Auto-GPT: Bringing AI Agent Development to New Heights 28011.13.2 Prompt_Chainer 28111.13.3 Auto_Chain 28211.13.4 AgentGPT: Unleashing the Power of Autonomous AI Agents 28211.13.5 BabyAGI: A Glimpse Into the Future of Task-Driven AI 28311.13.6 SimpleaiChat 28311.13.7 GradientJ: Building LLM-Powered Applications with Ease 28411.14 Conclusion 284References 28512 LangChain: Simplifying Development with Language Models 287Sangeetha Annam, Merry Saxena, Ujjwal Kaushik and Shikha Mittal12.1 Introduction 28812.2 Phases and Characteristics of LLM Application 28912.3 Components and Key Elements of LLM 29012.4 Types and Architecture of LLM 29312.5 Benefits and Approaches of LLM 29612.6 Building an LLM Application 29912.7 Use Cases 300References 30213 Addressing Ethical Challenges in LLMs: Bias and Misinformation 305Pummy Dhiman and Amandeep Kaur13.1 Introduction 30513.2 LLM Evolution Tree 30813.2.1 Bert 30913.2.2 Gpt 31113.3 Types of LLMs 31313.4 Limitations of LLMs 31413.5 Factors Contributing to Bias and Misinformation Generation 31613.6 Methods to Address Bias and Misinformation 31713.7 Conclusion 319References 320Part 5: LLM-Powered Applications 32314 LegalEase: Application Development with LangChain Framework 325Nidhi Malik, Lakshita Chhikara, Abhilakshay and Ambika Thakur14.1 Introduction 32514.1.1 Large Language Model 32614.1.2 General Architecture 32714.1.3 Examples of LLMs 32914.1.4 Benefits 32914.1.5 Industry Applications 33014.2 LangChain 33114.2.1 Key Features of LangChain 33114.2.2 Key Components 33314.2.3 Who Should Explore 33514.3 Example of Application Development 33514.3.1 Key Features 33614.3.2 Purpose and Benefits 33614.4 Development Steps 33714.4.1 Libraries and Imports 33714.4.2 Environment Setup 34014.4.3 Data Collection 34114.4.4 User Interface Setup 34214.4.5 Document Summarization 34314.4.6 Querying the Document 35514.5 Conclusion 362References 36315 Unveiling the Potential of Massive Language Models in Software Engineering: Exploring Opportunities, Addressing Risks, and Comprehending Implications 365Mitali Chugh15.1 Introduction 36615.2 Harnessing the Power: Abilities of Large Language Models 36715.3 Navigating Challenges: Risks and Ethical Considerations 36915.4 Ethical Application: Strategies and Frameworks 37115.5 Establishing Ethical Frameworks for Accountability 37215.6 Collaborative Standards: Industry and Research Collaboration 37315.7 Transformative Effects: Broader Implications in Software Engineering 37515.8 Shaping the Future: Prospective Directions of Large Language Models 37715.9 Conclusion 378References 37916 Multidimensional Impacts of Generative AI and an In-Depth Analysis of LLMs with Their Expanding Horizons in Technology and Society 383Rubal Jeet, Mohammed Wasim Bhatt, Maher Ali Rusho, Aadam Quraishi and Mahesh Manchanda16.1 Introduction 38416.2 Literature Review 38616.3 Proposed Methodology 38916.4 Results 40216.5 Conclusion 408References 409Part 6: Responsible AI 41317 Responsible AI: Ethical Considerations in Generative AI 415Kamal Kumar and Poonam17.1 Introduction 41617.1.1 Defining Generative AI 41617.1.2 Distinguishing Machine Learning Approaches 41717.1.3 Brief History and Recent Breakthroughs 41717.1.4 Overview of Key Generative Architectures and Techniques 42017.1.4.1 Autoregressive Models 42017.1.4.2 Generative Adversarial Networks (GANs) 42017.1.4.3 VariationalAutoencoders (VAEs) 42017.1.4.4 Diffusion Models 42117.1.4.5 Self-Supervised, Meta and Multi-Task Learning 42217.1.5 Promising Applications and Benefits 42217.2 Key Ethical Considerations, Risks, and Challenges 42317.2.1 Societal Biases and Unfair Representational Harms 42317.2.2 Truth Manipulation and Attribution Difficulties 42417.2.3 Violations of Consent, Privacy, and Agency 42417.2.4 Misuse Potentials Across Fraud, Deceit, and Sabotage 42417.2.5 Broader Societal Impacts on Economics, Culture and Psychology 42517.3 Guiding Principles and Frameworks for Responsible Generative AI 42517.3.1 Transparency 42617.3.2 Justice, Fairness, and Inclusion 42617.3.3 Non-Maleficence 42617.3.4 Responsibility and Accountability 42617.3.5 Privacy and Data Protection 42617.4 Governance Strategies for Trustworthy Generative AI Innovation 42717.4.1 AI Ethics Guidelines and Organizational Policies 42717.4.2 Laws, Regulations, and Dynamic Governance Complexities 42717.4.3 Technical Approaches to Fairness, Transparency and Control 42717.4.4 Stakeholder Participation and Public Discourse Ethics 42817.5 Recommendations for Key Generative AI Stakeholders 42817.5.1 Guidelines for Technology Researchers and Developers 42817.5.2 Strategies for Organizations, Platforms, and Corporations 42917.5.3 Ethical Governance Strategies for Organizations 42917.5.4 Policy Options for Governments and Lawmakers 42917.5.5 Priorities for Broader Industry Governance Entities 43017.5.6 Considerations for Civil Society Groups, Activists, and General Public 43017.5.7 The Impact of Generative AI Like ChatGPT on Education 430Significant Risks and Difficulties to Surmount 431Research Priorities for the Future 43117.6 Conclusions 432References 43318 From Prototyping to Deployment: Human-Centered Design Practices in Responsible AI Innovation 435Jyoti Snehi, Manish Snehi, Isha Kansal and Vikas Khullar18.1 Introduction 43618.2 Literature Review 441Overview of Human-Centered Design Principles 443Responsible AI 447Gaps in Existing Research 451Methodology 452Research Design 452Rationale for Qualitative Approach 452Human-Centered Design in AI Prototyping 456Distinctions and Issues 456User Research and Personas 456Early-Phase Prototyping 457Iterative Design and Feedback Loops 457Ethical Considerations in AI Prototyping 458Identifying Ethical Challenges 458Incorporating Ethical Guidelines Into Prototyping 458Case Studies of Ethical AI Prototyping 459From Prototyping to Development 459Transitioning From Prototype to Full Development 460Ensuring Consistency in HCD Practices 460Collaboration Across Multidisciplinary Teams 461Tools and Techniques for Managing Development Phases 461Human-Centered Design in AI Deployment 462Challenges and Solutions 463Common Challenges in Implementing HCD in AI 463Solutions and Best Practices 465Lessons Learned From Case Studies 467Framework for Human-Centered and Responsible AI 46918.3 Conclusion 471References 47219 Toward Accurate Abbreviation Disambiguation in Medical Texts: A Comparative Study of AI Models 475A. Pandey and M. Saini19.1 Introduction 47619.2 Related Work 47719.3 Datasets 47919.4 Methodology 48019.4.1 Data Collection 48119.4.2 Pre-Processing 48119.4.3 Vector Feature Extraction 48219.4.4 Classification Model 48419.5 Results and Discussion 48819.6 Conclusion 491References 491Index 495