Unit–1- Revisiting AI Project Cycle & Ethical
- 1. 1 Unit–1: Revisiting AI Project Cycle & Ethical Frameworks for AI Project Cycle Project Cycle – Project cycle refers to life cycle of any project that describes different stages of project, with each stage is separated by one another and meeting certain objective. AI Project Cycle is about the process of converting real life problem to computer based AI model. Five Stages of AI Project Cycle - 1. Problem Scoping (1st Stage of AI Project Cycle) 2. Data Acquisition (2nd Stage of AI Project Cycle) 3. Data Exploration (3rd Stage of AI Project Cycle) 4. Modeling (4th Stage of AI Project Cycle) 5. Evaluation (5th Stage of AI Project Cycle) 1. Problem Scoping (1st Stage of AI Project Cycle) Problem scoping is to focus on the problem for which we are aiming. It is the process through which the problem is being defined and a detailed review of the problem is gathered for analysis. 4W’s Canvas / Problem Composition In order to solve a problem, the problem statement must be clearly defined. It is very important to know that the problem is affecting whom, what is the cause, in which situation problem may be faced and why should it be solved? 4Ws canvas also known as Problem Statement Template- The problem statement template is used to frame 4W’s into a paragraph to describe the problem, the stakeholder involved, the area affected and how solving the problem would benefit them. The Problem Statement refer to a clearly defined problem that states the factors behind the problem (the nature of problem), why problem is to be solved and for whom the solution will be beneficial (the stakeholder) This method consist of collectively answering question in four categories - 1. Who 2. What 3. Where 4.Why 1. Who Block - Help to find who are getting affected directly or indirectly due to the problem under analysis 2. What Block - In this stage we find the nature of the problem. This stage tells – What is the problem and how can we know that it is problem? 3. Where Block - This block helps to look into the situation in which problem arises or the context of the problem and the area where the problem is prominent. 4. Why Block - In this block we think about- Why we want to solve the problem. 2. Data Acquisition (2nd Stage of AI Project Cycle) Data acquisition is process of collecting data from various sources for the purpose of analytical operation like training and prediction. Data acquisition is also known as data collection.
- 2. 2 Characteristics of quality data – Data plays an important role in success of an AI project. Quality data has these characteristics – i. Relevance – Data must be related to scoped problem. ii. Accuracy – Data must be correct in every detail. iii. Completeness – Data must be consisting of every information required by the selected AI project. iv. Timeliness – Information must be up to date. v. Reliability – Data must be collected from reliable sources. vi. Validity – Data must be according to requirement. Type of data used in AI Project Data used in AI are broadly categories in two groups - i. Structured Data/Labeled Data - Structured data is that data that has purposely designed, pre-defined structure as per existing data model. II. Unstructured Data/ Unlabeled Data - Unstructured data is unprocessed and is generally generated by machine-led system which may be not fit for the some well-defined system. Some data acquisition sources: - i. Survey - Survey is a way of collecting information by asking questions directly to the customers. - Survey can collect either qualitative or quantitative data or both. ii. Web Scraping - Web data extraction (also known as web scraping, web harvesting, screen scraping) is a technique of extracting huge amount of data from websites on the internet using a web browser. iii. Sensors - Sensors often called Transducers, converts real world phenomenon like temperature, force, movement in to voltage or current signals that can be used as input. iv. Cameras - Camera can use to collect data in the form of image and videos. Live data can be acquired from web camera and CCTV. v. Observation - Observation means careful and systematic viewing of facts as they occur. - Observation done by a. Studying collective behavior and social situations. b. Following up of individual element of the situation. c. Understanding the situation in their interrelation. d. Getting the details of the situation. vi. Application Program Interface (API) - API helps to collect data from other application. API help to connect one application to another, that’s why we are able to use object or content of one application in another application 3. Data Exploration (3rd Stage of AI Project Cycle) Data Exploration is a process of arranging the collected data in to a form which can be analyzed or through which we can derive useful information. In Data Exploration process redundant (अनावश्यक ) or unrequired data are removed, missing values are handled and then characteristics of data are thoroughly understood.
- 3. 3 Data Visualisation – Data Visualisation refers to the process of representing data visually or graphically, by using visual elements like charts, graph, diagrams and map. Some visualisation tools - 1. Scatter Chart 2. Bubble Chart 3. Line Graph 4. Pie Chart 5. Bar Graph 6. Histogram 7. Heat Map - In this type of graphical representation, value is represented by different colours. Ex. – map showing population density. 8. Timeline – It an be use to depict historical events, critical milestones of a project. Use to represent data against time. 9. Node Link Diagram – Use to show interconnected things through the use of nodes and link lines. 10. Word Cloud – Represent the frequency of word. The frequency of word determines its priority. The word with highest priority will be drawn wit larger font size. Common software use foe Data Visualisation Most commonly use software for data visualization are MS-Excel, Google chart, Tableau 4. Modeling (4th Stage of AI Project Cycle) Modeling is a process in which AI-Enabled algorithms are being designed as per the requirements of the system and AI model is trained using the collected data repeatedly until it starts producing desired result. Categories of AI Models - AI model is classified in two categories – i. Model Driven ( Rule Based ) Rule Based AI model is also known as Model-Driven AI model refers to the branch of AI where models are developed using the algorithms having pre-defined labels, rules patterns and relationships. The Rule Base AI is used when we have known or labeled dataset. ii. Data Driven ( Learning Based ) Learning Based AI Model is also known as Data-Driven AI model, refers to branch of AI where models are trained to learn by lot of data. Here are no pre-defined patters, rules, relationships or algorithms. The Learning Based approach is used when data is unknown, random or unlabeled. 5. Evaluation (5th Stage of AI Project Cycle Evaluation is one of the crucial phase in the cycle of AI project. It ensures the reliability and fairness of a model. It access how well AI model performs its intended task. Primary objective of Evaluation id to access the AI model’s performance, identifying its strengths and weaknesses. Confuse Matrix is one of the way of evaluating AI mode. Terms associated with AI Evaluation – i. True Positive (TP) – Model predict the output correct for positive result. ii. False Positive (TP) – Model predict the output wrong for positive result. iii. True Negative (TP) – Model predict the output correct for negative result. iv. False Negative (TP) – Model predict the output wrong for negative result.
- 4. 4 Method of Evaluation 1. Accuracy- It is measure of correct prediction (Positive and Negative both) Accuracy = 𝑇𝑃+𝑇𝑁 𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁 2. Precision (Positive Predictive Value) – Precision is also known as Positive Predictive Value. It is ration of true positive prediction to total positive prediction Precision = 𝑇𝑃 𝑇𝑃+𝐹𝑃 3. Recall/Sensitivity – It is measure the portion of actual positive instance that are correctly identified. Recall = 𝑇𝑃 𝑇𝑃+𝐹𝑁 4. F1 Score – It is a harmonic mean between Precision and Recall which provide balance between two matrics. Introduction to AI Domains Artificial Intelligence becomes intelligent according to the training it gets. For training, the machine is fed with datasets. According to the applications for which the AI algorithm is being developed, the data fed into it changes. With respect to the type of data fed in the AI model, AI models can be broadly categorized into three domains: 1. Statistical Data Statistical Data is a domain of AI related to data systems and processes, in which the system collects numerous data, maintains data sets and derives meaning/sense out of them. Ex.- Price Comparison Websites (PriceGrabber, PriceRunner, Shopzilla, DealTime, PolicyBazar) These websites are being driven by lots and lots of data. If you have ever used these websites, you would know, the convenience of comparing the price of a product from multiple vendors in one place. 2. Computer Vision Computer Vision that depicts the capability of a machine to get and analyse visual information and afterwards predict some decisions about it. The entire process involves image acquiring, screening, analysing, identifying and extracting information. This extensive processing helps computers to understand any visual content and act on it accordingly. In computer vision, Input to machines can be photographs, videos and pictures from thermal or infrared sensors, indicators and different sources. Examples of Computer Vision i. Agricultural Monitoring Computer vision is employed in agriculture for crop monitoring, pest detection, and yield estimation. Drones with cameras capture aerial images of farmland, which are then analysed to assess crop health and optimize farming practices. ii. Surveillance Systems Computer vision is used in surveillance systems to monitor public spaces, buildings, and borders. It can detect suspicious activities, track individuals or vehicles, and provide real-time alerts to security personnel.
- 5. 5 3. Natural Language Processing Natural Language Processing, abbreviated as NLP, is a branch of artificial intelligence that deals with the interaction between computers and humans using the natural language. Natural language refers to language that is spoken and written by people, and natural language processing (NLP) attempts to extract information from the spoken and written word using algorithms. The ultimate objective of NLP is to read, decipher, understand, and make sense of human languages in a valuable manner. Examples of Natural Language Processing i. Email filters Email filters are one of the most basic and initial applications of NLP online. It started with spam filters, uncovering certain words or phrases that signal a spam message. ii. Machine Translation NLP is used in machine translation systems like Google Translate and Microsoft Translator to automatically translate text from one language to another. These systems analyze the structure and semantics of sentences in the source language and generate equivalent translations in the target language. AI Ethical AI Ethics refers to moral principles, guidelines and value that governs the development, deployment and use of artificial intelligence technologies. Pillars of AI Ethics i. Accountability ii. Transparency iii. Fairness iv. Reliability v. Privacy vi. Safety and security vii. Sustainability Ethical Frameworks for AI Frameworks - Frameworks are a set of steps that help us in solving problems. It provides a step-by-step guide for solving problems in an organized manner. - Frameworks are step-by-step guidance on solving problems. - Frameworks offer a structured approach to problem-solving, ensuring that all relevant factors and considerations are taken into account. - Additionally, they serve as a common language for communication and collaboration, facilitating the sharing of best practices and promoting consistency in problem- solving methodologies. Ethical Frameworks Ethics are a set of values or morals which help us separate right from wrong. Hence, - Ethical frameworks are frameworks which help us ensure that the choices we make do not cause unintended harm. - Ethical frameworks provide a systematic approach to navigating complex moral dilemmas by considering various ethical principles and perspectives.
- 6. 6 By utilizing ethical frameworks, individuals and organizations can make well informed decisions that align with their values and promote positive outcomes for all stakeholders involved. Need of Ethical Frameworks for AI Bias could result in unwanted outcomes in AI solutions. Think of the hiring algorithm which was biased against women applicants! AI is essentially being used as a decision-making/ influencing tool. As such we need to ensure that AI makes morally acceptable recommendations. Ethical frameworks ensure that AI makes morally acceptable choices. If we use ethical frameworks while building our AI solutions, we can avoid unintended outcomes, even before they take place. Types of Ethical Frameworks The various types of ethical frameworks are classified as follows: 1. Sector-based Frameworks 2. Value-based Frameworks 1. Sector-based Frameworks: These are frameworks tailored to specific sectors or industries. In the context of AI, one common sector-based framework is Bioethics, which focuses on ethical considerations in healthcare. It addresses issues such as patient privacy, data security, and the ethical use of AI in medical decision-making. Sector-based ethical frameworks may also apply to domains such as finance, education, transportation, agriculture, governance, and law enforcement. Bioethics Bioethics is an ethical framework used in healthcare and life sciences. It deals with ethical issues related to health, medicine, and biological sciences, ensuring that AI applications in healthcare adhere to ethical standards and considerations. Principals of Bioethics – i. Autonomy (Respect for individual) Patients have right know their medical condition, AI should assist in medical decision without overriding patient consent. ii. Beneficence (Doing good) AI application should aim to improve patient outcome and well-being. AI should enhance diagnosis accuracy and medical research advancements. Ethical Frameworks Ethical Frameworks Ethical Frameworks Value-based Frameworks Sector-based Frameworks Virtue-based Rights-based Bioethics Utility-based
- 7. 7 iii. Non-Maleficence (Do no harm) AI must minimize the risk such as biased algorithms, incorrect diagnoses or privacy breaches. iv. Justice (Fairness and equality) Ai should ensure equal access to healthcare services, avoiding biases in data or decision making. 2. Value-based Frameworks: Value-based frameworks focus on fundamental ethical principles and values guiding decision-making. It reflects the different moral philosophies that inform ethical reasoning. Value-based frameworks are concerned with assessing the moral worth of actions and guiding ethical behavior. They can be further classified into three categories: i. Rights-based: Prioritizes the protection of human rights and dignity, valuing human life over other considerations. It emphasizes the importance of respecting individual autonomy, dignity, and freedoms. In the context of AI, this could involve ensuring that AI systems do not violate human rights or discriminate against certain groups. ii. Utility-based: Evaluates actions based on the principle of maximizing utility or overall good, aiming to achieve outcomes that offer the greatest benefit and minimize harm. It seeks to maximize overall utility or benefit for the greatest number of people. In AI, this might involve weighing the potential benefits of AI applications against the risks they pose to society, such as job displacement or privacy concerns. iii. Virtue-based: This framework focuses on the character and intentions of the individuals involved in decision-making. It asks whether the actions of individuals or organizations align with virtuous principles such as honesty, compassion, and integrity. In the context of AI, virtue ethics could involve considering whether developers, users, and regulators uphold ethical values throughout the AI lifecycle. Ethical Issues under Ethical Framework i. Transparency – It enable AI system explainable, understandable and accountable ii. Privacy – It is about how data collected, processed and stored. iii. Fairness and Non-Discrimination – Avoiding bias and discrimination in decision making. iv. Security – The negligent security can have wide-range impact. v. Accountability – For the action of AI who will be accountable – the machine, algorithm, developer or owner. The lack of accountability can lead to unaddressed injustice and diminished trust in AI technology. vi. Misinformation – Misinformation significant effect the public opinion. vii. Job Displacement/Unemployment – Automation will replace certain job roles. viii. Deep fake – It become serious ethical concern. It can override security measures as it can by-pass voice and facial recognition.