Artificial Intelligence Project Cycle and Ethics

1. What are the five stages of the AI Project Cycle in the correct order?

Problem Scoping → Data Acquisition → Data Exploration → Modelling → Evaluation

Data Acquisition → Problem Scoping → Modelling → Data Exploration → Evaluation

Problem Scoping → Modelling → Data Acquisition → Data Exploration → Evaluation

Data Exploration → Problem Scoping → Data Acquisition → Evaluation → Modelling

The AI Project Cycle begins with Problem Scoping, where the specific issue to be addressed is defined. Next, Data Acquisition involves gathering relevant data necessary for analysis. Following this, Data Exploration allows for understanding and preprocessing the data to identify patterns and insights. The Modelling stage involves selecting and training algorithms to create predictive models. Finally, Evaluation assesses the model's performance and effectiveness in solving the initial problem, ensuring that the project meets its objectives. This sequence is crucial for a structured and successful AI project.

Explanation

The AI Project Cycle begins with Problem Scoping, where the specific issue to be addressed is defined. Next, Data Acquisition involves gathering relevant data necessary for analysis. Following this, Data Exploration allows for understanding and preprocessing the data to identify patterns and insights. The Modelling stage involves selecting and training algorithms to create predictive models. Finally, Evaluation assesses the model's performance and effectiveness in solving the initial problem, ensuring that the project meets its objectives. This sequence is crucial for a structured and successful AI project.

2. Which type of AI works on fixed rules and does NOT learn from data?

Learning-based AI

Rule-based AI

Reinforcement Learning AI

Deep Learning AI

Rule-based AI operates on a predefined set of rules and logic, making decisions based on these established guidelines without the ability to adapt or learn from new data. This contrasts with other types of AI, such as learning-based or deep learning systems, which improve their performance by analyzing data and adjusting their algorithms over time. Rule-based systems are particularly useful in environments where consistent, predictable outcomes are required, as they follow strict protocols without deviation.

Explanation

Rule-based AI operates on a predefined set of rules and logic, making decisions based on these established guidelines without the ability to adapt or learn from new data. This contrasts with other types of AI, such as learning-based or deep learning systems, which improve their performance by analyzing data and adjusting their algorithms over time. Rule-based systems are particularly useful in environments where consistent, predictable outcomes are required, as they follow strict protocols without deviation.

3. Match each type of Machine Learning with its correct example.

Supervised Learning

Unsupervised Learning

Reinforcement Learning

Deep Learning

Submit

4. Which of the following are the five principles of AI Ethics?

Fairness

Reliability & Safety

Profitability

Privacy & Security

Inclusiveness

Transparency

The five principles of AI Ethics focus on ensuring that AI systems operate in a manner that is just and equitable (Fairness), dependable and secure (Reliability & Safety), respect individual rights (Privacy & Security), promote diversity and accessibility (Inclusiveness), and provide clarity about their operations and decisions (Transparency). These principles collectively aim to foster trust and accountability in AI technologies, ensuring they benefit society while minimizing potential harms. Profitability is not considered an ethical principle in this context, as it does not directly address the ethical implications of AI deployment.

Explanation

The five principles of AI Ethics focus on ensuring that AI systems operate in a manner that is just and equitable (Fairness), dependable and secure (Reliability & Safety), respect individual rights (Privacy & Security), promote diversity and accessibility (Inclusiveness), and provide clarity about their operations and decisions (Transparency). These principles collectively aim to foster trust and accountability in AI technologies, ensuring they benefit society while minimizing potential harms. Profitability is not considered an ethical principle in this context, as it does not directly address the ethical implications of AI deployment.

Submit

5. In the 4W Problem Canvas used in Problem Scoping, what does the 'W' in 'Where' refer to?

Who is responsible for solving the problem

Where the problem occurs

What resources are needed

Why the data is collected

In the 4W Problem Canvas, 'Where' specifically addresses the location or context in which the problem manifests. Understanding 'Where the problem occurs' is crucial for identifying the environment and circumstances surrounding the issue, which helps in accurately assessing its impact and formulating effective solutions. By pinpointing the geographical or situational aspects of the problem, stakeholders can tailor their approaches to address the specific needs and challenges present in that setting.

Explanation

In the 4W Problem Canvas, 'Where' specifically addresses the location or context in which the problem manifests. Understanding 'Where the problem occurs' is crucial for identifying the environment and circumstances surrounding the issue, which helps in accurately assessing its impact and formulating effective solutions. By pinpointing the geographical or situational aspects of the problem, stakeholders can tailor their approaches to address the specific needs and challenges present in that setting.

6. A model predicts that an email is spam, but it is actually not spam. This is called a ____.

A false positive occurs when a model incorrectly identifies a non-spam email as spam. In this scenario, the model's prediction misclassifies a legitimate email, leading to potential negative consequences, such as missing important communications. False positives are significant in classification tasks as they reflect the model's inaccuracies, highlighting the need for improved algorithms to minimize such errors and enhance overall performance.

Explanation

A false positive occurs when a model incorrectly identifies a non-spam email as spam. In this scenario, the model's prediction misclassifies a legitimate email, leading to potential negative consequences, such as missing important communications. False positives are significant in classification tasks as they reflect the model's inaccuracies, highlighting the need for improved algorithms to minimize such errors and enhance overall performance.

Submit

7. Deep Learning (DL) is a subset of Machine Learning (ML), which is itself a subset of Artificial Intelligence (AI).

True

False

Deep Learning (DL) refers to advanced algorithms that mimic the human brain's neural networks, enabling machines to learn from vast amounts of data. It is a specialized area within Machine Learning (ML), which encompasses a broader range of techniques for training algorithms to recognize patterns and make decisions. Machine Learning, in turn, is a subset of Artificial Intelligence (AI), the overarching field focused on creating systems that can perform tasks typically requiring human intelligence. This hierarchical relationship illustrates how DL is an advanced form of ML, which is part of the larger AI domain.

Explanation

Deep Learning (DL) refers to advanced algorithms that mimic the human brain's neural networks, enabling machines to learn from vast amounts of data. It is a specialized area within Machine Learning (ML), which encompasses a broader range of techniques for training algorithms to recognize patterns and make decisions. Machine Learning, in turn, is a subset of Artificial Intelligence (AI), the overarching field focused on creating systems that can perform tasks typically requiring human intelligence. This hierarchical relationship illustrates how DL is an advanced form of ML, which is part of the larger AI domain.