Gen AI: Foundational Concepts & Landscape

What is a Gen AI Agent?

An application that achieves goals by observing the world, reasoning, and acting using tools and models.

What is the reasoning loop?

The agent’s process of observing, interpreting, reasoning, and acting (often with prompt engineering).

What are agent tools?

Functionalities allowing agents to interact with the environment — processing data, accessing APIs, etc.

What is the model in an AI agent?

The "brains" — algorithms that learn from data to make predictions or generate content.

What is the Model Garden in Vertex AI?

A collection of pre-trained models from Google, third parties, or open-source providers.

What is Model Builder in Vertex AI?

A tool for training your own models either from scratch (custom ML frameworks) or using AutoML.

What is AI infrastructure?

The core hardware (e.g., GPUs, TPUs) and software used to train, store, and run AI models.

What is Edge AI?

Running AI models on devices or local servers close to where the data is generated.

What is LiteRT?

A tool from Google for deploying AI models to edge devices efficiently.

What is Gemini Nano?

Google’s smallest and most efficient AI model, optimized for on-device performance.

What are key Needs to consider before starting a Gen AI project?

• Scale

• Customization

• User interaction

• Privacy

• Latency

• Connectivity

What are key Resources to consider before starting a Gen AI project?

• People (expertise)

• Money (budget)

• Time (timeline)

What is Artificial Intelligence (AI)?

The field of building machines that can perform tasks typically requiring human intelligence, such as learning and decision-making.

What is Machine Learning (ML)?

A subfield of AI where machines learn from data to perform specific tasks.

What is Deep Learning?

A subset of ML that uses neural networks with many layers to learn complex patterns in data.

What are Foundation Models?

Large ML models trained on massive amounts of unlabeled data, allowing them to generalize well across many tasks.

What are Large Language Models (LLMs)?

Foundation models specifically designed to understand and generate human language.

What is structured data?

Data organized into a searchable format, often stored in relational databases.

What defines high-quality data?

Data that is accurate, complete, consistent, and relevant.

What does data accessibility mean in AI?

The extent to which data is available, usable, and in a format suitable for training models.

What are the stages of the ML lifecycle?

1. Data ingestion and preparation

2. Model training

3. Model deployment

4. Model management

What is Secure AI?

Practices that protect AI systems from threats and ensure safe deployment.

What is Ethical AI?

The practice of ensuring AI systems do not cause harm and are used fairly and responsibly.

What is SAIF?

Secure AI Framework — helps manage security risks in AI/ML models.

What is Supervised Learning?

Learning from labeled data to predict outcomes.

What is Unsupervised Learning?

Learning patterns and groupings from unlabeled data.

What is Reinforcement Learning?

Learning by interacting with the environment, receiving rewards and penalties.

What is a Gen AI Agent?

An application that achieves goals by observing the world, reasoning, and acting using tools and models.

What is the reasoning loop?

The agent’s process of observing, interpreting, reasoning, and acting (often with prompt engineering).

What are agent tools?

Functionalities allowing agents to interact with the environment — processing data, accessing APIs, etc.

What is the model in an AI agent?

The "brains" — algorithms that learn from data to make predictions or generate content.

What is Vertex AI?

Google Cloud’s unified ML platform for building, deploying, and managing ML and Gen AI solutions.

What is the Model Garden in Vertex AI?

A collection of pre-trained models from Google, third parties, or open-source providers.

What is Model Builder in Vertex AI?

A tool for training your own models either from scratch (custom ML frameworks) or using AutoML.

What is AI infrastructure?

The core hardware (e.g., GPUs, TPUs) and software used to train, store, and run AI models.

What is Edge AI?

Running AI models on devices or local servers close to where the data is generated.

What is LiteRT?

A tool from Google for deploying AI models to edge devices efficiently.

What is Gemini Nano?

Google’s smallest and most efficient AI model, optimized for on-device performance.

What are key Needs to consider before starting a Gen AI project?

• Scale

• Customization

• User interaction

• Privacy

• Latency

• Connectivity

What are key Resources to consider before starting a Gen AI project?

• People (expertise)

• Money (budget)

• Time (timeline)