ad
ad

How to use WatsonX.ai and integrate with Watson assistant

Education

Introduction

Introduction

Hey there! Welcome to a captivating session of technical talk with Brijesh. I am Brijesh, and in this session, we're going to dive deep into WatsonX and generative AI features. But before we jump in, let's have a quick look at what AI and generative AI are.

What is AI?

Artificial Intelligence (AI) uses a rule-based approach, where explicit instructions and predefined rules are programmed to enable the system to perform tasks. These rules are designed by human experts based on their understanding of a problem domain.

What is Generative AI?

Generative AI, on the other hand, uses neural networks to identify patterns and structures within existing data to generate new and original data. It enables users to create new content such as images, text, sound, animations, 3D models, and other types of data.

Example of Generative AI

Imagine you are teaching a child. If the child doesn't understand something, you change your approach until he captures and understands the information based on your pattern of explanation. This is how generative AI works. It adapts and generates new content based on learned patterns.

Types of Artificial Intelligence

  1. ANI (Artificial Narrow Intelligence)
  2. AGI (Artificial General Intelligence)
  3. ASI (Artificial Super Intelligence)

Though we won't cover these in detail in this session, it’s good to know the types.

WatsonX Overview

WatsonX is an enterprise-ready AI and data platform designed for businesses. It provides three main products:

  1. WatsonX.ai: A studio for New Foundation models, generative AI, and machine learning where you can train, validate, fine-tune, and deploy your AI models.
  2. WatsonX.data: Storing and analytics using a purpose-fit, open lake house architecture.
  3. WatsonX.governance: A toolkit that accelerates AI workflows built with responsibility, transparency, and explainability.

Why Choose WatsonX?

  1. Open Technology: Provides a variety of models to cover enterprise use cases and compliance requirements.
  2. Trusted: Transparent, responsible, and governed models pre-trained on trusted datasets.
  3. Targeted: Designed for enterprises with multiple models tailored for particular domains like finance, banking, or e-commerce.
  4. Empowering: Allows you to train, fine-tune, deploy, and govern your AI models, giving you complete ownership of the created value.

How to Use WatsonX

Training with Example

To train your model:

  1. Create a project and prepare data either by connecting to various data sources or uploading files directly.
  2. Use the prompt lab to test your data and find the best suitable model from available foundation models like Google Flan UL 20B.

Generative AI

Generative AI models use neural networks to identify patterns and structures within the data to generate new and original content. For example, if you input a set of questions and answers, the model will learn to generate answers based on those patterns.

Using Watson Assistant

Here’s how you can integrate WatsonX with Watson Assistant:

  1. Create a New Project: In Watson Assistant, create a new project and define your user flow.
  2. Integrate WatsonX: In the integration section, add a custom extension for WatsonX and provide necessary API details.
  3. Create Actions: Define actions that the assistant should take when interacting with users.
  4. Set up Variables: Create necessary variables for storing user input and responses from WatsonX.
  5. Publish and Deploy: Publish the assistant and deploy it on your website or application using the provided script.

Testing the Integration

Test the integration by asking questions and verifying if you receive appropriate responses from the WatsonX model. If it’s not working as expected, make adjustments in the variables, prompts, and models to fine-tune the answer accuracy.

Keywords

FAQ

Q1: What is the difference between traditional AI and generative AI?

A1: Traditional AI relies on rule-based approaches with predefined instructions, while generative AI uses neural networks to identify patterns and generate new data based on those patterns.

Q2: What are the main WatsonX products?

A2: The main WatsonX products are WatsonX.ai, WatsonX.data, and WatsonX.governance.

Q3: How do I integrate WatsonX with Watson Assistant?

A3: You integrate WatsonX with Watson Assistant by adding a custom extension in the integration section, setting up necessary variables, defining actions, and then publishing and deploying the assistant.

Q4: Can WatsonX be used on-premises?

A4: Yes, WatsonX is available both in the cloud and on-premises, ideal for customers with strict compliance rules or who prefer to work with confidential data within their infrastructure.

Q5: How do I choose the best model for my use case in WatsonX?

A5: Use the prompt lab to test different foundation models with your data, and select the model that provides the most accurate responses for your use case.

This article should now provide a comprehensive understanding of how to use WatsonX and integrate it with Watson Assistant while addressing some common queries and summarizing the essential points.