Brainscape's Knowledge GenomeTM


Top Flashcards (Certified)
All Flashcards

AIGP > Section 3: Understanding the Foundations of AI Governance > Flashcards

Section 3: Understanding the Foundations of AI Governance Flashcards

(58 cards)

Start Studying
1
Q

What are the four shared features of AI definitions?

A

Technology, Autonomy, Human involvement, Output.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is Inference in machine learning?

A

The machine learning model’s output, such as a decision or prediction.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What are the four categories of machine learning models?

A

Supervised, Unsupervised, Semi-supervised, Reinforcement.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What are the Supervised learning techniques?

A

Classification, Regression, Support Vector Machines (SVM), Support Vector Regression (SVR).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What are the two categories of unsupervised learning?

A

Clustering and Association rule learning.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is Transfer learning?

A

An algorithm learns one task and then applies this knowledge to a different but related task.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is Broad AI?

A

A subset of artificial narrow intelligence involving multiple narrow AIs working together.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What are the three components of an expert system?

A

Knowledge base, Inference engine, User interface.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What two things does Fuzzy logic rely on?

A

Linguistic variables and Fuzzy rules.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What are the four steps of a fuzzy logic system?

A

Fuzzification, Rule evaluation, Aggregation, Defuzzification.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is a Parameter in machine learning?

A

An internal variable the model learns from training data and adjusts during training.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is a Variable?

A

A measurable attribute that can take on different values; either numerical/quantitative or categorical/qualitative.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What are Features in machine learning?

A

Input variables or attributes used to make predictions; characteristics the model learns from.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are the Seven different AI use cases?

A

Recommendation, Recognition, Detection, Forecasting, Goal-driven optimization, Interaction support, Personalization.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What are the four components of the AI tech stack?

A

Compute, Storage, Network, Software.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What does Compute include?

A

CPU, GPU, Memory, Storage, and Data processing.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What is the difference between FLOPS and FLOPs?

A

FLOPS = Speed at a single point in time (hardware).
FLOPs = Total computational work done over time (model).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What are the three types of compute?

A

Serverless, High-performance compute (HPC), Trusted Execution Environments.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What are the four stages of data storage?

A

Ingestion, Preparation, Training, Output.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

What is Transmission Control Protocol (TCP)?

A

A protocol that describes how data is transferred across the network between devices.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

What are Platforms in the context of AI?

A

Software to plan, design, develop, implement, and deploy AI systems.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

What is Data post-processing?

A

Adjusting model outputs after production to improve fairness or meet business requirements.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

What are the Challenges with data labeling?

A

Low quality data, Lack of quality assurance, Lack of manual scaling capabilities.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

What is Data observability?

Study These Flashcards
A

Monitoring the health of the data ecosystem using pre-determined metrics.

25
What is Validation data used for?
To assess ongoing model performance during training, fine-tune parameters, and prevent overfitting.
26
What is Underfitting?
When a model fails to capture the complexity of data due to too few parameters or features.
27
What are the Engines of Growth (the nine to memorize)?
Mobile, Metaverse, Cloud computing, Computer vision, Augmented/Virtual reality, Internet of Things, Privacy-enhancing technologies, Social media, Blockchain.
28
What are the three harms individuals may suffer?
Civil rights, Economic opportunity, Safety.
29
What are the Privacy-specific harms (6)?
Aggregation, Incorporation into training data, Inference, Secondary use, Lack of transparency, Inaccuracy.
30
What are the Economic harms?
Job displacement, Bias, Failure to reach certain groups.
31
What are Group harms?
Facial recognition errors, Mass surveillance, Civil rights and socio-economic divides.
32
What are Harms to society?
Democratic processes, Trust in institutions, Access to public services, Employment, Disinformation, Misinformation.
33
What are Harms to companies?
Reputational, Cultural, Economic, Legal/Regulatory, Acceleration risks.
34
What are the Alignment Goals?
Intended, Specified, Emergent.
35
What are the Two types of misalignment?
Inner misalignment, Outer misalignment.
36
What are the Types of bias?
Algorithmic, Computational, Cognitive, Societal, Implicit, Sampling, Temporal.
37
What are the four categories of AI risks?
Security, Operational, Privacy, Business.
38
What are some attack types in AI?
Model inversion, Model extraction/theft, Model evasion, Malicious algorithm.
39
What is Data persistence?
When data outlives the data subject; requires data deletion and retention policies.
40
What is Spillover data?
Incidental data collection tangential to intended data collection.
41
What is a Harms taxonomy?
A map of harms to anticipate risks, implement controls, and track regulatory needs.
42
What is MITREs Panoptic?
A privacy threat assessment framework based on context and actions in attacks.
43
What is the Ryan Calo framework?
A harms framework focused on perception of unwanted observation (subjective harms) and forced data use (objective harms).
44
Who are Daniel Solove and Danielle Citron, and what did they categorize?
Privacy harms into categories: physical, reputational, relationship, economic, discrimination, psychological, failure to inform, lack of control, chilling effects.
45
What are the Operational risks?
Hardware, Storage, High-speed network, Expertise, Environment.
46
What are the Five themes of the Sociotechnical Harms Taxonomy?
Representational, Allocative, Quality-of-service, Interpersonal, Social/societal.
47
What are the Four elements of the CSET AI Harm Taxonomy?
An entity experienced a harm event linked to a consequence of an AI system.
48
What is Interpretability?
The ability to explain an AI system’s reasoning before an output is provided
49
What are the Four main characteristics of Trustworthy AI?
Human-centric, Accountable, Transparent, Acts in a legal and fair manner.
50
How do you operationalize Trustworthy AI?
Leadership buy-in, Create RAI principles, Understand AI business purpose, Embed into risk framework, Develop technical standards, Update structures, Ensure human oversight.
51
What factors must be considered when tailoring AI governance (6)?
Organization size, Maturity, Industry/Sector, Products/Services, Strategic Objectives, Risk Tolerance.
52
What are the Developer responsibilities?
Collect required data, Document provenance, Understand policy requirements, Gather user feedback.
53
What are the Deployer responsibilities?
Understand objectives, Provide transparency, Ensure responsible deployment.
54
What are the User responsibilities?
Understand acceptable use policy, Know system’s business purpose, Report incidents and provide feedback.
55
What are the components of Building a governance framework?
Principles, Risk tolerance, Industry/Sector, Jurisdiction, Ability to implement, AI’s purpose and relationship to business.
56
What are the AI governance principles (7)?
Pro-innovation mindset, Consensus-driven, Outcome focused, Law/Tech/Industry-agnostic, Non-prescriptive, Risk-centric, End-to-end accountability.
57
What steps can be taken to achieve Leadership buy-in for AI governance?
Identify an AI champion, Demonstrate RAI as a differentiator, Proactively identify challenges and solutions.
58
What are the Components of an AI strategy (6)?
Understand business objectives and needs, Assess data governance maturity, Develop ethical framework, Choose technologies and tools, Prioritize AI skills, Gain leadership and employee buy-in.

AIGP (4 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2025 Bold Learning Solutions. Terms and Conditions