Ch 1 AI Project Lifecycle Class 8 Notes Important Points
Ch 1 AI Project Lifecycle Class 8 Notes
Introduction
AI is a type of technology that gives machines and computers the ability to perform work that normally requires human intelligence. These machines can perform complex tasks done by humans, like analysing data, finding patterns, predicting trends, solving problems, and making decisions by using data and learning from past experiences.
In simple words, AI is human-made intelligence that allows machines to behave intelligently.
Functions of Artificial Intelligence
1. Analyse data: AI studies large amounts of data to find meaning and make decisions. It can manage huge data quickly and efficiently.
2. Recognize patterns: AI recognises patterns and relationships in data, such as trends, similarities, or repeated behaviours.
3. Learn from experience: AI keeps improving over time by learning from new data. The more it learns, the better it performs.
4. Make predictions or decisions: AI can predict outcomes or make decisions, often based on probabilities and past data.
Uses of Artificial Intelligence in Daily Life
Artificial Intelligence (AI) is becoming an important part of our everyday lives. It is used in many fields like Healthcare, Home etc. to make our tasks easier.
Various uses of AI in everyday life are:
In Healthcare: AI helps doctors diagnose diseases faster and more accurately by looking at medical scans and reports.
At Home: AI powers smart home devices like speakers that play your favourite music when you tell them to play.
Getting Around: AI is the main technology behind self-driving cars and helps navigation apps find the quickest routes.
On Your Smartphone: AI is used when you use a voice assistant on your phone to set a reminder or when your photo app automatically puts pictures of the same person together.
Personalised Apps: It recommends new videos and products you might like based on what you have done in the past.
How does AI learn?
The real magic behind Artificial Intelligence lies in data. Artificial intelligence and Machine Learning systems learn from data. These models learn by analysing huge datasets and recognising patterns. Every AI-driven program works correctly when supported by accurate and reliable data.
What is an AI Project?
An AI project is a step-by-step process of developing a system or model using artificial intelligence technology to solve a real-world problem by learning from past data. The entire process of building, testing, and improving the system step by step is called the AI project lifecycle.
AI Project lifecycle:
The AI project lifecycle is an iterative, six-stage framework, for developing, deploying, and maintaining AI tools.
Key Phases of AI Project lifecycle:
- Define the problem,
- Data Collection and Preparation for modelling,
- AI Model Development and training
- Model evaluation and its refinement,
- Deployment of the model
- Continuous Monitoring and Maintenance.
1. Define the problem:
The first and most crucial step in an AI project lifecycle is to define or identify the problem. At this stage, we must determine how the system should work and whether the AI system is really required for this project.
Let us take an example of an Automatic car wash system
In this problem, the steps are:
- The car enters the bay
- Sensor detects Vehicle Position
- The car moves forward on the conveyor
- For a fixed duration, water is sprayed
- Soap is applied for a fixed time
- Brushes rotate at programmed speed
- Rinse cycle run
- Dryer activates
- The system stops when the car exits
In this example, there is no learning or decision-making beyond the programmed logic. No matter how big or small the car is, the machine always does the same things in the same way. It doesn’t think, this car is really dirty; maybe I should scrub it longer. It simply does what it is told to do.
These kinds of systems have fixed set rules and do not learn from data. This is an example of automation, not AI.
In AI-based car wash cameras with AI system scan the car to detect its size, dirt level, and type of dirt, and the system then decides water spray time, soap usage, and brush speed based on the dirt instead of fixed timing.
Let us take another example of an Automatic Lift System
In this problem, the steps are:
- User presses call button
- Lift moves to the requested floor
- Door opens automatically
- User enters and selects destination floor
- Door closes
- Lift moves to selected floor
- Door opens for exit
- Door closes and system resets
In AI-based lift systems, sensors and intelligent algorithms analyze passenger demand, traffic patterns, and usage behavior in real time. The system does not simply follow fixed rules—it makes decisions to optimize waiting time and efficiency.
Some other examples:
Traffic signal control:
Time-based (automation): Traffic lights change after every 30 seconds, no matter how many cars are waiting.
AI-based: Traffic lights AI models are designed to adjust green light timing based on traffic density.
Irrigation system:
Time-based (automation): This irrigation system has a timer that waters the plants every day at 7 AM, no matter what the weather is like.
AI-based: The irrigation system AI models are designed to water plants based on soil moisture and weather using predictive analytics.
2. Data Collection and Preparation:
Data collection and preparation are the next important phase of an AI project lifecycle. AI systems mainly learn from past data, so the quality of that data is a crucial factor affecting their performance.
The AI will be bad if the data is bad. In the world of AI, there is a well-known saying: Garbage In, Garbage Out.
Some of the sources of data collection are:
Sensors: It is a device that measures physical changes in the environment and convert them into a measurable electrical signal. For example, temperature sensor measures the heat, or light sensor can detect the light intensity and convert it into an electrical signal.
Surveys: Data is collected by asking questions and collecting responses from people.
Websites: Data can be collected from online sources such as articles, social media platforms, and websites.
Historical records: These include past data stored in files, reports, or records.
Let us take an example:
To predict whether a student will score more than 75% marks in the exam or not. For this problem, we can collect the following data from the previous students:
1. Attendance
2. Hours studied per day
3. Previous test marks
4. Participation in class Yes or No
5. Final exams result
Data Preparation:
For developing an AI system, we need a large amount of data. However, during the data collection, the data we usually collect is raw or unorganised.
It may contain errors, missing values, or unnecessary information. Therefore, we should clean and organise the data before using it to train an AI model.
The following steps are commonly used during the data preparation:
A. Data Cleaning:
- Remove errors or irrelevant data
- Fill missing values
- Remove duplicate data
B. Data Formatting:
- Convert text into number (if required)
- Arrange data in a spreadsheet neatly
C. Data Labelling:
- Assign correct categories (e.g., Spam/Not Spam, Pass/Fail)
3. Model Development and Training:
Model development means creating an AI model that solves the problem defined in stage 1. An AI model works like the brain of the system. It learns from data and uses that learning to make predictions.
Let us look at the previous example whether a student will score more than 75% or not. Create an AI model to address this issue by using the following data to learn from past students:
- Study hours
- Attendance
- Test marks
- Participation in class
This learning process of a model is known as model training.
The model may notice that students with attendance above 75% usually score well. Or that students who study less than 2 hours rarely cross 75%. It’s like the model is connecting the dots!
Model Evaluation and Refinement
After training, the model is tested using new data to check how well it works. Testing means checking the model using a test dataset. A test dataset is a set of data that was not used during training.
Let us consider from our training, the model learned that a student would get more than 75% if they have:
- More than 80% attendance
- Study more than two hours
- Get more than 75 on tests
- Participate in class
Now, this trained model will work for two new students, student 4 and student 5. Based on what it learned, the model predicts that student 4 will get more than 75%, but student 5 will get less than 75%.
The model correctly predicts for 2 out of 2 students, so the accuracy of the developed model is 100%.
There may be a condition: if our model correctly predicts only 2 out of 10 students, then its accuracy
is only 20%. In this case, model refinement is required.
Model Refinement
Model refinement means improving the model when it does not give satisfactory results.
We can improve the model by making changes as given below:
- Adding more data
- Removing incorrect data
- Changing model settings
- Training the model again
This process continues until the model performs well.
5. Model Deployment:
After the model has been trained, tested, and refined, the next stage is to deploy it in real-world use. In simple words we can say that the model is now ready for use :
- Take the input data (e.g., images, signals, or voice)
- Make prediction
- Help users in the real situation
Real-life examples of deployment:
- Spam detection in email apps
- Recommendation system in shopping apps
- Voice assistant
- Face unlock in smartphones
6. Monitoring and Maintenance
Monitoring is the continuous checking of the performance of the deployed AI model whereas Maintenance is to update the model to keep it accurate and efficient over time. It regularly checks the following essential tasks:
- Is the system working correctly?
- Is the model giving the correct prediction?
- Has the accuracy remained the same or decreased?
If the deployed model is monitored regularly, it can help to identify when it is not working properly or when its accuracy reduces. In such cases, the model needs to be updated.
The model can be updated by:
- Adding new data
- Retraining the model
- Fixing the errors
- Improving performance
Important Points to Remember
- AI systems learn from data.
- A clear problem must be defined before building an AI system.
- AI models identify patterns in data.
- AI projects follow a step-by-step lifecycle.
- AI systems improve over time with more data