UX for Enterprise ChatGPT Solutions A Practical Guide to Designing Enterprise-Grade LLMs
Create engaging AI experiences by mastering ChatGPT for business and leveraging user interface design practices, research methods, prompt engineering, the feeding lifecycle, and more Key Features Learn in-demand design thinking and user research techniques applicable to all conversational AI platfor...
| Otros Autores: | , |
|---|---|
| Formato: | Libro electrónico |
| Idioma: | Inglés |
| Publicado: |
Birmingham, England :
Packt Publishing
[2024]
|
| Edición: | First edition |
| Materias: | |
| Ver en Biblioteca Universitat Ramon Llull: | https://discovery.url.edu/permalink/34CSUC_URL/1im36ta/alma991009849117006719 |
Tabla de Contenidos:
- Cover
- Title page
- Copyright and credits
- Dedication
- Foreword
- Acknowledgment
- Contributors
- Preface
- Table of Contents
- Part 1: UX Foundation for Enterprise ChatGPT
- Chapter 1: Recognizing the Power of Design in ChatGPT
- Technical requirements
- Approach 1 - The no-code approach
- Approach 2 - code with Node.JS, Python, or curl
- Traversing the history of conversational AI
- The importance of UX design for ChatGPT
- Understanding the science and art of UX design
- The science of design
- The art of design
- It takes a village to create superb UX
- Setting up a customized model
- Summary
- References
- Chapter 2: User Research
- Surveying UX research methods
- Understanding user needs analysis
- Surveys for conversational AI
- Survey checklist
- Case study on an effective survey
- Designing insightful interviews
- Defining research objectives
- Selecting participants
- Develop a structured interview program
- Pilot the interview process and program
- Conduct the structured interviews
- Record and document findings
- Data analysis
- Report findings
- Summary of the interview process
- Getting started with conversational analysis
- Tagging a log file should focus on each interaction
- Define success and failure categories
- Trying conversational analysis
- Exploring the examples from the case study
- Generate enhancements and bugs from groups of issues
- Score results
- Results
- Summary
- References
- Chapter 3: Identifying Optimal Use Cases for ChatGPT
- Understanding use case basics
- Use case or user stories
- Establishing a baseline with ChatGPT
- Example use case for a ChatGPT instance - patching software
- Creating a user story from a use case
- Prioritizing ChatGPT opportunities from the use case
- Aligning LLMs with user goals
- Applications of ChatGPT.
- Examples of generative AI outside of chat
- Avoiding ChatGPT limitations, biases, and inappropriate responses
- Lack of real-time information
- Complex or specialized topics
- Long-form content generation
- Long-term memory
- Sensitive information
- Biased thinking
- Emotion and empathy
- Ethical and moral guidance
- Critical decision making
- Programming and debugging
- Translation accuracy
- Educational substitution
- Don't force-fit a solution
- Summary
- References
- Chapter 4: Scoring Stories
- Prioritizing the backlog
- WSJF
- User Needs Scoring
- Scoring enterprise solutions
- Examples of scoring
- Putting a backlog into order
- Patching case study revisited
- Extending tracking tools with scoring
- Try the User Needs Scoring method
- Creating more complex scoring methods
- Working with multiple backlogs in Agile
- Real-world hiccups with scoring
- I know Agile, and this is not WSJF
- The use of simple numbers one to four
- Weighting factors
- Severity seems complicated to judge
- The cost is so high that we can't ever get the work done
- Grouping issues into bugs to protect the quality
- How to work WSJF into the organization
- Summary
- References
- Chapter 5: Defining the Desired Experience
- Designing chat experiences
- Chat-only experiences
- Integrating ChatGPT into an existing chat experience
- Enabling components for a chat experience
- Designing hybrid UI/chat experiences
- Chat window size and location
- Tables
- Forms
- Charts
- Graphics and images
- Buttons, menus, and choice lists
- Links
- Creating voice-only experiences
- Designing a recommender and behind-the-scenes experiences
- Overarching considerations
- Accessibility
- Internationalization
- Trust
- Security
- Summary
- References
- Part 2: Designing
- Chapter 6: Gathering Data - Content is King.
- What is in a ChatGPT foundational model
- Incorporating enterprise data using RAG
- Understanding RAG
- Limitations of ChatGPT and RAG
- Building a demo with enterprise data
- Cleaning data
- Other considerations for creating a quality data pipeline
- Resources for RAG
- Community resources
- Summary
- References
- Chapter 7: Prompt Engineering
- Giving context through prompt engineering
- Prompt 101
- Designing instructions
- Basic strategies
- Quick tricks to always keep in mind
- A/B testing
- Prompt engineering techniques
- Self-consistency
- General knowledge prompting
- Prompt chaining
- Program-aided language models
- Few-shot prompting
- Andrew Ng's agentic approach
- Reflection
- Tool use
- Planning
- Multi-agent collaboration
- Advanced techniques
- Summary
- References
- Chapter 8: Fine-Tuning
- Fine-tuning 101
- Prompt engineering or fine-tuning? Where to spend resources
- Token costs do matter
- Creating fine-tuned models
- Fine-tuning for style and tone
- Using the fine-tuned model
- Fine-tuning for structuring output
- Generating data should still need a check and balance
- Fine-tuning for function and tool calling
- Fine-tuning tips
- Wove case study, continued
- Prompt engineering
- Fine-Tuning for Wove
- Summary
- References
- Part 3: Care and Feeding
- Chapter 9: Guidelines and Heuristics
- Applying guidelines to design
- Adapting heuristic analysis for conversational UIs
- 1 - Visibility of system status
- 2 - Match between a system and the real world
- 3 - User control and freedom
- 4 - Consistency and standards
- 5 - Error prevention
- 6 - Recognition rather than recall
- 7 - Flexibility and efficiency of use
- 8 - Aesthetic and minimalist design
- 9 - Help users recognize, diagnose, and recover from errors
- 10 - Help and documentation.
- Is there an 11th possible heuristic?
- Building conversational guidelines
- Web guidelines
- A sample guideline set for hybrid chat/GUI experiences
- Some specific style and tone guidelines with examples
- Flow order can reduce interactions
- Case study
- Handling errors - repair and disfluencies
- Summary
- References
- Chapter 10: Monitoring and Evaluation
- Evaluate using RAGAs
- The RAGAs process
- Synthesizing data
- Evaluation metrics
- User experience metrics
- Other metrics
- Monitoring and classifying the types of hallucination errors
- OpenAI's case study on quality and how to measure it
- Systematic testing processes
- Testing matrix approach
- Improving retrieval
- The wide range of LLM evaluation metrics
- Monitor with usability metrics
- Net Promoter Score (NPS)
- SUS
- Refine with heuristic evaluation
- Summary
- References
- Chapter 11: Process
- Incorporating design thinking into development
- Find a sponsor
- Find the right tools and integrate Generative AI
- Be religious… at first
- Avoid "unknown unknowns"
- Always evolve and improve
- Agile does not mean "no requirements"
- Team composition and location matters
- Manage Work in Progress (WIP) and technical debt
- Focus on customer value
- Incorporate the design process into the dev process
- Designing a content improvement life cycle
- Inputs for conversational AIs
- Inputs for recommender UIs
- Inputs for backend AIs
- Monitoring Monday
- Analysis Tuesday (and Wednesday's workup)
- Treatment Thursday and fault-finding Friday
- What doesn't fit into a week is still important
- Conclusion
- References
- Chapter 12: Conclusion
- Applying learnings to the new frontier
- Double-checking what feels right
- Set clear goals
- Know your processes
- Know the data
- Align and be accountable
- Prioritize thoughtfully.
- Automate with intention
- Building processes that fit the solution
- Wrapping up the journey
- References
- Index
- Other Books You May Enjoy.
