کتاب طراحی رابط‌های کاربری هوش مصنوعی

درک مدل‌های زبانی بزرگ و سیستم‌ها

Early Predictive Language Models for Consumers • A Short History of AI • The Beginning: Symbols • User Experience in the AI Era • Know Your User • Technical and Evaluatory Skills • Ethical Considerations in AI Design • Communication: Knowledge Sharing and Role Clarity • Reading This Book: The Input-Computation-Output Structure • Inputs • Computation • Outputs • Summary

قابلیت، کشف و ارکستراسیون

Capabilities: What Can the Model Do? • Training • Evals • Discovery • Intent with AI • Discoverability and Initialization • Adoption and Momentum Behavior • Orchestration • Models, Agents, and Tool Selection • Usage and Metering • Permissions and Capabilities • Library Management and Assets • Summary

طراحی برای ورودی‌های هوش مصنوعی

From Multics to Google Search: The Shifting Burden • Three Channels of Intent • Implicit Context • Explicit Prompting • The CARE Framework • Prompt Intent • Risks in Explicit Prompting • Direct Manipulation • Input Modalities • Text • Image • Voice • Guiding Users Toward Better Inputs • Starter Prompts as Product Positioning • Summary

محاسبات: طراحی برای مرحله پردازش و تولید

The Computation Pipeline: From Input to Output • Step 1: Input Processing and Preparation • Step 2: Routing • Step 3: Generation and Inference • Don’t Make Me Wait: Managing Latency and Delays • Visual Methods for Mitigating Latency • Position: Where the Action Is At • Speed: Difficult to Predict • Notify Me: Notifications and Status Indicators • The Value of Segmenting by User Need • When the System Doesn’t Deliver: Designing for Errors • Errors in AI • Error Messages • Midstream Errors and Retrieval • Maintaining Momentum • Summary

خروجی: طراحی تحویل و ارائه پاسخ‌های مدل زبانی بزرگ

Designing Outputs: The Output Is Not the Answer • Design Principles of Outputs • Designing for Clarity • Designing for Verifiability • Designing Grounding • Designing for Actionability • Designing for Adjustability • Designing for Multiturn Outputs • Watermarking and Detection • Understanding AI Misuse • Risks of Generative AI Content • How Watermarking Works • Image and Video Watermarking • Detection Systems and the Future of Watermarking • Managing Problematic Outputs • Summary

هوش مصنوعی عاملی: طراحی برای سیستم‌هایی که برنامه‌ریزی، اقدام و سازگاری می‌کنند

Understanding Agentic Design Patterns • The Reflection Pattern: Building in Self-Critique • How Reflection Works in Practice • Design Challenges with Reflection • Reflection Across Different Domains • The Tool Use Pattern: From Adviser to Operator • The Shift from Static to Dynamic Assistance • Design Implications of Tool Use • The Planning Pattern: Breaking Down Complexity • Designing for Visible Planning • Conditional Planning and Branching • The Multiagent Collaboration Pattern: Specialized Coordination • Multiagent Interface Challenges • Conflict Resolution in Multiagent Systems • The ReAct Pattern: Adaptive Problem Solving • Implementation and Use Cases • Designing for Iterative Agents • Moving from Engineering to Experience • Where Users Experience Agentic Patterns • Processing • Clarification • Progress Communication • Alternative Progress Patterns: Browser and Computer Use Visualization • Checkpoints • New Principles for Agentic Interface Design • Reveal the Plan • Prioritize What Matters Most • Design for Shared Control • Summary