Topic1-Types of AI and how can we earn by using these different types of AI.

 

In this lecture we would learn about the Types of AI and how can we earn by using these different types of AI.

We will understand AI by making a story in our mind about different tools of AI.

There are lot of fears about Artificial Intelligence, that it would replace our jobs. But reality is contrary. It is lack of interest in AI and sticking on conventional methods that will terminate your current jobs.

There are lot of visions of AI. AI is specific implications for each job.

AI concept is that we are trying to do with the help computer that a man can do.

TWO ASPECTS OF AI

General AI- For away concept

Narrow AI-

Basic human thinking is being automated. 90% of us could not apprehend how rapidly AI is becoming the part of our life. Minimum 10 to 20 tools are launched each morning.

STAGES OF LEARNING AI

1.     Skill Development

2.     Assignments / Tools / Projects will be given to both physical and online students through email.

3.     Problem Solving. It is pertinent to mention that AI is not coding but coding is the tiny part of AI.

4.     Active Participation. Whole world is in race. Learning AI will help to create an effective self-image after learning from us. This learning will help to spread the knowledge at distant places and resulting spreading it rapidly.

5.     Work Ethics.

6.     Evaluation.

7.     Quiz.

8.     Presentations. It would be divided in two portions. Monthly at the end of each month and at the end of session.

It is difficult to apprehend that how much time our journey from Narrow AI to Generalized AI would take. Probably it can take one or two decades. No body knows.

Fast transformation from Chat GPT 4 to Chat GPT 5 have created alert for the developers of Chat GPT and they are showing their concerns on this fast transformation and asking for legislations to control the uncontrollable.

Those who have non-technical background should not leave the course and must stick to the course as we have lot of things for them to learn.

AI concepts would be develop

ed in your mind by testing the database and building the story in you mind for coming 4 Months.

We can elaborate this by using the examples from daily life.

Conflict is that AI and Modals are useless without data. It means that answers to lot of questions would not be known to Chat GPT. For Example, if we ask about table diffusion it would be unknown to Chat GPT.

Definition:

Explicit Programming means conventional way without coding we have to teach computation. We us such methods by which computers learns without coding.

WHY AI IS APPEALING AND VALUED CAREER OPTION

It is estimated that AI would create 13 trillion dollars value by the year 2030.

It must be kept in mind that our scope is to use Developed AI not be a Machine Learning engineering or Scientist. Example is like driving a bike and not be an Engineering involved in manufacturing a bike.

MOST POPULAR WAY TO DO AI

Machine learning (ML) is a type of artificial intelligence (AI) that allows software applications to become more accurate at predicting outcomes without being explicitly programmed to do so. Machine learning algorithms use historical data as input to predict new output values.

Machine Learning is branch of AI which focus on methods than can learn from examples. It learn from previous data available.

DIFFERENCE BETWEEN MACHINE LEARNING AND CLASSICAL AI

DATA + ALOGRITHM + TRAINING = MODAL

AVAILABLE DATA 10,000
Known 7000	Unknown 3000

 PROMPT ENGINEERING

Prompt engineering or prompting is an AI (artificial intelligence) concept of how to talk to an AI system like ChatGPT and get a desired response. For example, instead of giving the AI a simple prompt like "define a computer," you could add additional information to the prompt to get a better result. In other words, you could give a prompt like "define a computer in one paragraph as if I was five years old" and get more of a result of what you want.

Prompt Engineering itself going to be big field.

Certain Image How many prompts

Only logic and courage is required for prompt Engineering.

Medical students can do lot by using prompt engineering.

The journey from Chat Gpt 3 to Chat Gpt 4 has not be made by giving more data but making it more smart by using computing and Algorithms.

Chapt GPT 4 has been developed by trainings using data based in Chat GPT 3.5

CONVENTIONAL MACHINE LEARNING

Conventional Machine learning are meant equation of match. We convert input data features form of math to more accurate to more smart modal.

Algorithm

a process or set of rules to be followed in calculations or other problem-solving operations, especially by a computer.

"a basic algorithm for division"

 

An example from daily life on making a bread was discussed. It was told when we follow the steps required to make the bread, result is definitely bread.

GPT has also algorithm.

Difference between Modal and Algorithm was discussed. It was told that data plus Algorithm will result Modal. Suppose when we train Algorithm with data it becomes a trained modal.

. Types of AI like ML generative AI were discussed

A question from class was asked that has every modal has same algorithm. It was answered that No, we have lot of algorithms that would result different trained modals with the help of data.

It was discussed that what if no data is available? Would No Modals would be trained. It was told that generative modals produce new data points from the data on the basis of modal is trained. Example already given about mechanic was discussed. It was told that if we have trained the mechanic about the 04 different types of engines, if an engine that is similar to any one of the 04 trained engines come before the mechanic, the mechanic will apply already learnt knowledge on this similar engine on the basis of knowledge / training already learnt.

SUPERVISED LEARNING:

Some body helping algorithm to define whether it is working right or wrong.

Suppose there is project of 100 images and the modal have to select a cat image out of these. If you give a non-cat image to the modal and modal responds that it is cat image. You will give feed back to the Modal that you have given the wrong answer.

In Supervised Learning the trainer is aware about the picture and its name. This type of data is called Labelled data. Labelled data means that data is in pairs of input and output. When we allocate a number / name to the picture, it is called labelling.

   Above examples were explained.

In short supervised learning there is labelled data means that input and out put of the data is known to the trainer.

There is limitation of Supervised Learning. What if all the data in thousands / millions available with us could not be labelled.

TYPES OF SUPERVISED LEARNING

There are two types of Supervised Learning

1.     Classification

2.     Regression

 In classification we have defined classes. For example, in example given above we have three defined classes of Cat, Dog and Chicken and there is no more class. If we have 1000 pictures these pictures can either be Cat or Dog or Chicken.

If I am predicting the score of Babar Azam based on his score in last ten innings or I am trying to predict the value of my house, this is called regression.

Labelled data is used for Classification and Score is used for Regression

Following Question from online learners were asked

1.     What is Supervised Learning

2.     What is API Purpose is to stop the audience from sensitive information.

3.     What is difference between Classification and Regression

4.     How can we learn Prompt Engineering.

A source Andrew NG free curse on web https://www.deeplearning.ai/ was referred by Sir Irfan.

Structured and Unstructured Data

Data in organized form like in excel is called Structured Data

Data in raw form like speech is called Unstructured data

5.     Data in ChatGpt has no reference, should be rely on it?

6.     What will happen if data given to the Modal is not accurate? Answer was garbage in garbage out. Solution is that we will test the data before feeding it. Even data would be validated after feeding. To tackle this problem there is full job that is called Data Engineering who assures the accurate data and useful to develop a modal.

7.     ​Can we create a API which can collect data from both chatgpt and brad at the same time

Answer was that it should not be blindly relied upon.