Chapter 19 – AI working with and for humans

Contents

19.1  Overview
19.2  Introduction
19.3  Levels and Types of Human Contact
19.3.1  Social Scale
19.3.2  Visibility and Embodiment
19.3.3  Intentionality
19.3.4  Who Is in Control
19.3.5  Levels of Automation
19.4  On a Device -- Intelligent User Interfaces
19.4.1  Low-level Input
19.4.2  Conversational User Interfaces
19.4.3  Predicting What Next
19.4.4  Finding and Managing Information
19.4.5  Helping with Tasks
19.4.6  Adaptation and Personalisation
19.4.7  Going Small
19.5  In the World -- Smart Environments
19.5.1  Configuration
19.5.2  Sensor Fusion
19.5.3  Context and Activity
19.5.4  Designing for Uncertainty in Sensor-rich Smart Environments
19.5.5  Dealing with Hiddenness -- A Central Heating Controller
19.6  Designing for AI--Human Interaction
19.6.1  Appropriate Intelligence -- Soft Failure
19.6.2  Feedback -- Error Detection and Repair
19.6.3  Decisions and Suggestions
19.6.4  Case Study: onCue -- Appropriate Intelligence by Design
19.7  Towards Human--Machine Synergy
19.7.1  Tuning AI Algorithms for Interaction
19.7.2  Tuning Interaction for AI
19.8  Summary

Glossary items referenced in this chapter

accuracy, activity recognition, agents!embodied, algorithmic accountability, appropriate intelligence, appropriation, auto-complete, automated decision, autonomous car, base rate, Bayes Theorem, boundary objects, camera!frame, chatbot, ChatGPT, classification, Clippy, computer vision, confidence, confidence in output, confidence measures, confusion matrix, conservative warnings, context of human activity, context-aware interaction, convolutional neural network, data detector, decision tree, decision tree in IVR, deep neural network, deterministic ground, dot-com period, dynamic function allocation, EAGER, echo chambers, ELIZA, emergent behaviour, epistemic interaction, expert system, explainable AI, explicit confirmation, false negative, false positive, feature interaction, fourier analysis, function allocation, gesture recognition, gig economy, GPT-3, handwriting recognition, heuristic evaluation function, Hick's law, home automation, Human Computer Interaction, human-in-the-loop, hybrid AI, hypertext, implicit feedback, implicit interaction, incidental interaction, industrial robotics, intelligent user interfaces, intentionality, interactive voice response, internet of things, interpretation of user actions, liberal suggestions, machine learning, Microsoft, modal user interface, multimodal interfaces, named entity recognition, natural language processing, neural network, onCue, optimal, optimal route, pattern matching, pattern recognition, pen-based systems, personal information management, Phone Slave, privacy, probability, probability estimate, programming-by-demonstration, Query-by-Browsing, reCAPTCHA, recommender systems, redundancy, relevance feedback, responsibility, robotics, ROC, sample rate, search engine, search terms, semantic desktop, semantic web, semi-autonomous car, sensor fusion, sensor redundancy, signal processing, smart dust, social media, speech recognition, statistical techniques, synergistic interaction, threshold, transfer learning, transformer model, transparency, transpose, trigger, uncertainty, uncertainty of sensors, user interface, voice assistants