Index
A
Accountability, AI ethics
Accuracy
Actuator
Advanced home theater systems
Affected users, AI system
Aggregation
AI capabilities
AI products
classification and categorization
CV
input-output mappings
knowledge representation
natural language processing (NLP)
pattern recognition
perception, motion planning and control
prediction
ranking
recommendation
speech and audio processing
trust in
AI ethics
accountability and regulation
beneficial AI
beneficial machine
bias
black box model
control problem
ethics-based design
explainable AI
human in the loop
Independent Review Committees
law
liability
manipulation
privacy and data collection
safe AI
security
transparency
trustworthy AI
AI explainability
AI metrics
accuracy
confusion matrix
precision
precision vs. recall tradeoff
recall
AI Opportunities
jobs
break down a job
implications
running trainer
tasks
mapping user journeys
SeeMapping user journeys
AI personality
cultural norms
grammatical person
guidelines
inclusivity, strive for
risks of personification
tone of voice
Airbnb
AI system
affected users
AI output, control over
calibrating trust, plan for
cause-and-effect relationship
confidence-based explanations
control
control mechanisms, types of
data control
data use explanations
decision-makers
description
designing, guidelines for
editability
experimentation, explaining through
explanation
explanations, types of
functionality
global controls
hand off gracefully
internal stakeholders
no explanation
optimization
opting out
plant recognizer
regulators
removal and reset
set expectations for adaptation
transparency
AI techniques
backpropagation
DL
SeeDeep learning (DL)
GANs
knowledge graphs
reinforcement learning
supervised learning
SeeSupervised learning
transfer learning
unsupervised learning
Alan Turing
Algorithmic bias
Altruism
Amazon
Artificial general intelligence (AGI)
Artificial intelligence (AI)
acting humanly
AGI
building products
definitions
examples-based approach
feedback loops in
foundational disciplines
business case
computer engineering
control theory
cybernetics
economics
linguistics
mathematics
neuroscience
philosophy
psychology
intelligent everything
narrow AI
rational agent approach
rational thinking
rules-based approach
separate field
SAIs
STT
substrate independent
thinking humanly
user experience
Audio entity recognition
B
Backpropagation
Balance control and automation
Beneficial AI
Beneficial machine
Benevolence
Bias
causes of
facial recognition
reducing
team representation, lack of
training data
Big Red button
Black box model
Blameless postmortem
Building AI
approaches
examples-based approach
products
rule-based approach
C
Calibration, trust
Cardiogram
Cat detector
Categorization
Cause-and-effect relationship
Classification
Clustering
Cognitive division of labor
humans
machines
Collaboration
Common roles, AI product team
applied ML scientist
data engineer
data scientist
ML engineers
ML researchers
product designers
product managers
software engineers
Competence
Computer vision (CV)
Confidence-based explanations
categorical
data visualizations
guidelines
N-best
numeric
types of
Confusion matrix
Context errors
Control
feedback
global
hand off gracefully
types of
Control, AI systems
balance control and automation
guidelines
Cultural norms
D
Data
access
availability
compounding improvements
Data control
Data errors
Data requirements
Data use explanations
guidelines for
types of
Decision-makers
Deep learning (DL)
Descriptions
full explanations
guidelines
partial explanations
Desirability
Dual feedback
E
Editability
Effective collaboration
build user empathy
data requirements
don’t dictate
easy things are hard
encourage experimentation
functional requirements
hypothesis validation
motivated and engaged teams
product metrics
share problems, not solutions
user-centric narratives
Entity recognition
Error handling
AI, inevitable in
appropriate responses
disambiguation
“Errors” and “Failures
explanation, opportunities for
feedback, opportunities for
feedback, to finding new errors
graceful failure and handoff
guidelines for
humble machines
indication
intentional abuse
invisible errors
model errors
optimization
recalibrating trust
relevance errors
return control to user
stakes of the situation
strategies for
system errors
trust
type of error
user
user errors
user-perceived errors
Ethical implications
Ethics-based design
Everlaw
Examples-based explanation
Experimentation
Explainability
Explainability prototypes
Explainable AI
Explanations
confidence-based
data use explanations
guidelines for
types of
descriptions
internal assessment
qualitative methods
quantitative methods
types of
user validation
Explicit feedback
F
Facebook
Face recognition
Feature engineering
Feedback
adaptation, set expectations for
alignment
altruism
artificial intelligence, feedback loops in
changes, connection to
collaboration with team
collection
control
correction, error
dual feedback
during regular interactions, encouraging
editability
explicit feedback
guidelines
human-AI collaboration
ignore and dismiss
implicit feedback
limit disruptive changes
long-term response
make it easy
material rewards
on-the-spot response
opting out
removal and reset
reward
reward function
self-expression
social rewards
stakes of situation
symbolic rewards
timing and scope
transparency
types of
usage
user motivations
utility
Feedback loop
Fidelit
G
Generative adversarial networks (GANs)
Global controls
H
Hardware prototypes
Human + Machine
artificial collective intelligence
automating redundant tasks
Human-Machine Collaboration
improving machines
machine-machine collaboration
collective intelligence
addition
connections
improvement
missing middle
assistants
Cobots
managers
peers
tools
supermind
Human-AI Collaboration, feedback
Hyperparameters
I, J
Image classification
Image segmentation
Implicit feedback
Independent Review Committees, AI ethics
Information retrieval
Instacart’s partner app
Intelligent/rational agent
actuators
agent
description
environment
complex environments
simple environments
goals
sensors
Intentional abuse, error handling
Internal stakeholders, AI system
Invisible errors
background errors
happy accidents
K
Knowledge graphs
Knowledge representation
L
Law, AI ethics
Levels of Autonomy
Liability, AI ethics
Limit disruptive changes
Loading states and updates
Long-term response, feedback
M
Machine learning (ML)
depth
DL
feedback mechanism
reward function
risk of rewards
learning
mapping
probabilistic nature, AI
reinforcement learning
shallow learning
supervised learning
unsupervised learning
Machine translation
Manipulation
behavior control
personality
risks of personification
Mapping user journeys
AI benefits
categorization and classification
detecting anomalies
generate new data
identified jobs and tasks
natural language
personalization
prediction
ranking
recognition
recommendation
AI tasks
augmentation
automation
costs
data
quality improvements and gains
societal norms
time and effort
type of environment
SeeType of environment
experience mapping
journey mapping
service blueprints
story mapping
Material rewards
Minimum viable product (MVP)
fly on the wall
learning and validating assumptions
task completion
Mislabeled/misclassified data
Model errors
incorrect model
miscalibrated input
security flaws
Modern AI
Multiple AI techniques
Musical performance
N
Narrow AI
face recognition
machine translation
object detection
robotics
spam detection
speech to text
web search
Natural language processing (NLP)
entity recognition in text
information retrieval
machine translation
text classification
text entity recognition
N-best
Neural networks
O
Object detection
Objective function
Object tracking
Onboarding
One-on-one interviews
P
Painted door test
Pattern recognition
Perception
Personality and emotion
Precision
Predictability
Prediction
Privacy and data collection
ask permissions
data use
opting out
personally identifiable information, protection
regulations
terms and conditions
user data, protection
Problem-first approach
complete transparency
data being sparse
high speed optimization
limited information
low costs optimization
maintaining predictability
minimizing costly errors
peoples certain tasks
social intelligence
Product design discipline
Product designers
Product managers
Prototyping
benefits
building AI products
product development process
Prototyping AI products
building relevance prototypes
desirability
internal assessment
personalized experience
usability prototype
Q
Qualitative methods
R
Ranking algorithms
Raspberry Pi
Reach of data use
Recall
Recommendation
Redaction
Regression
Regulators, AI system
Reinforcement learning
Relevance errors
irrelevance
low-confidence
Reliability
Removal and reset, feedback
Rethinking Processes
Mercedes
Netflix
Reward
material rewards
social rewards
symbolic rewards
Reward function
feedback
Robotics
Robots
Amazon
robotic vacuum cleaner
self-driving vehicles
Zipline
S
Safe AI
Scope of data use
Search engine results page (SERP)
Security, AI ethics
Self-driving vehicles
Self-expression
Semantic network
Shallow learning
Smoke test
Social rewards
Societal norm
Speaker ID
Speech and audio processing
applications
audio entity recognition
speaker ID
speech synthesis
STT
wake word detection
Speech processing
Speech to text (STT)
Superintelligent AIs (SAIs)
Supervised learning
challenge
classification
labeled
regression
Support efficient dismissal
Symbolic rewards
System errors
data errors
incomplete data
mislabeled or misclassified data
missing data
T
Transfer learning
Transparency
Trigger/hot word
Trust
in AI
AI personality, guidelines
AI system
calibrating trust, plan for
cause-and-effect relationship
description
explanations, types of
internal stakeholders
transparency
benevolence
borrowing
building
calibration
competence
components of
control
errors
explainability
loading states and updates
onboarding
opportunities for building
personality and emotion
predictability
reliability
settings and preferences
user interactions
Trustworthy AI
Type of environment
continuous/discrete actions
dynamic or static environments
fully observable or partially observable
number of agents
predictable or unpredictable environments
time horizon
U
Unsupervised learning
Usability prototype
User-centric narratives
User errors
breaking user habits
unexpected/incorrect input
User interactions
cause-and-effect relationships
data use
set right expectations
User-perceived errors
context errors
failstates
User validation
fly on the wall
surveys and customer feedback
task completion
Utility feedback
V
Voice assistants
W, X, Y
Wake word
Wizard of Oz studies
Z
Zipline
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.