Unit 1: Introduction to Artificial Intelligence - Subjective Questions

Artificial Intelligence (AI) is the branch of computer science concerned with designing machines and software that can perform tasks normally requiring human intelligence.

The main objectives of AI are:

• Learning: Enabling systems to learn from data and experience.
• Reasoning: Allowing machines to draw conclusions and solve problems.
• Problem-solving: Finding suitable solutions to complex tasks.
• Perception: Interpreting inputs such as images, speech, and sensor data.
• Natural language processing: Understanding and generating human language.
• Planning and decision-making: Selecting appropriate actions to achieve a goal.
• Knowledge representation: Storing and organizing information so that it can be used effectively.

AI may use methods such as machine learning, deep learning, expert systems, search algorithms, and natural language processing.

In the context of AI, intelligence refers to the ability of an entity to perceive its environment, understand information, learn from experience, reason about alternatives, and take actions that help achieve desired goals.

Important characteristics of intelligence include:

• Adaptability: Adjusting behavior when circumstances change.
• Learning ability: Improving performance using experience or data.
• Reasoning: Applying rules or knowledge to reach conclusions.
• Creativity: Producing novel ideas or solutions.
• Decision-making: Selecting an appropriate action among alternatives.
• Communication: Exchanging information using language or other signals.

AI attempts to reproduce some or all of these capabilities in machines. However, machine intelligence is usually measured by performance on specific tasks, whereas human intelligence is broader and includes emotional, social, and common-sense abilities.

An intelligent system generally has the following characteristics:

• Perception: It receives information from the environment through sensors, cameras, microphones, or digital inputs.
• Knowledge representation: It stores facts, rules, relationships, or patterns in a usable form.
• Inference and reasoning: It processes available knowledge to derive conclusions.
• Learning: It improves its performance by analyzing examples, feedback, or past experience.
• Planning: It determines a sequence of actions required to reach a goal.
• Decision-making: It evaluates alternatives and chooses an action.
• Adaptation: It responds effectively to changing conditions.
• Autonomy: It can operate with limited human intervention.

For example, a self-driving vehicle perceives roads and obstacles, represents its surroundings, predicts possible events, plans a route, and makes driving decisions.

Weak AI and strong AI represent two different views of machine intelligence.

Weak AI can appear intelligent without actually understanding the task in a human sense. Strong AI remains a theoretical goal and raises important philosophical and ethical questions.

Most present-day AI systems are classified as weak AI, or narrow AI, because they are built to perform particular tasks within defined limits.

For example:

• A facial-recognition system identifies faces but cannot diagnose diseases.
• A translation system converts languages but may not understand the deeper meaning of a conversation.
• A recommendation system suggests products but cannot independently perform scientific research.

These systems may process large amounts of data and produce highly accurate results, but their abilities do not automatically transfer to unrelated tasks. They also do not possess confirmed self-awareness, general common sense, or human-like understanding. Therefore, even highly advanced systems remain narrow in purpose and are categorized as weak AI.

Strong AI, also called artificial general intelligence (AGI), refers to a machine that can understand, learn, reason, and apply knowledge across a wide range of tasks at a level comparable to humans.

Major challenges include:

• Generalization: Transferring knowledge from one situation to a completely different situation.
• Common-sense reasoning: Understanding everyday facts that humans rarely need to state explicitly.
• Contextual understanding: Interpreting meaning according to social, cultural, and situational context.
• Consciousness and self-awareness: Determining whether a machine can genuinely experience awareness.
• Learning efficiency: Humans often learn from a few examples, while many AI systems require large datasets.
• Robustness: Operating reliably in unfamiliar, uncertain, or changing environments.
• Ethical behavior: Making decisions consistent with human values and social responsibilities.

Strong AI has not been conclusively achieved and remains an important research objective.

Human intelligence and artificial intelligence have both similarities and significant differences.

Similarities:

• Both can process information and solve problems.
• Both can learn from examples or experience.
• Both can recognize patterns and make decisions.
• Both can use stored knowledge to perform tasks.

Differences:

• Humans possess emotions, consciousness, intuition, and biological needs; AI systems do not have these qualities in the human sense.
• Humans can often learn from a small number of examples, whereas AI frequently requires large datasets.
• Computers can perform calculations and search large databases much faster than humans.
• Human intelligence is generally flexible and transferable across domains, while most AI is specialized.
• Humans can use common sense and social understanding more naturally than current AI systems.

Thus, AI can exceed human performance in selected tasks but does not yet reproduce the full range of human intelligence.

The origins of AI can be traced to philosophy, mathematics, logic, and the development of computing machines.

• Ancient philosophers explored reasoning, knowledge, and the possibility of artificial beings.
• In the nineteenth and early twentieth centuries, formal logic and mathematical theories provided ways to represent reasoning.
• The development of programmable computers during the 1940s made automated calculation and symbolic processing possible.
• In 1950, Alan Turing proposed the question of whether machines could think and introduced the imitation game, later called the Turing Test.
• In 1956, the Dartmouth workshop formally introduced the term artificial intelligence. Researchers such as John McCarthy, Marvin Minsky, Allen Newell, and Herbert Simon contributed to the field.
• Early AI programs focused on theorem proving, game playing, symbolic reasoning, and problem-solving.

These developments established AI as an independent area of computer science.

Important milestones from 1950 to the 1970s include:

• 1950: Alan Turing published work asking whether machines could think and proposed the imitation game.
• 1956: The Dartmouth conference popularized the term artificial intelligence.
• Late 1950s: Researchers developed programs for symbolic reasoning, theorem proving, and game playing.
• 1958: John McCarthy developed the LISP programming language, which became important in AI research.
• 1960s: AI systems were created for problem-solving, natural language experiments, robotics, and medical consultation.
• 1966: ELIZA demonstrated early computer-based conversation using pattern matching.
• Late 1960s and early 1970s: Researchers developed knowledge-based systems and robots capable of planning in controlled environments.

Early progress created great optimism, but many systems worked only in limited environments and lacked sufficient computing power and real-world knowledge.

An AI winter is a period in which interest, funding, and confidence in AI decline after expectations are not achieved.

The main causes were:

• Excessive promises: Researchers and institutions predicted rapid achievement of human-level intelligence.
• Limited computing power: Computers lacked the memory and processing speed required for complex tasks.
• Insufficient data: Many AI methods could not perform well without large and varied datasets.
• Brittleness: Early systems performed well in restricted situations but failed when conditions changed.
• High development costs: Maintaining specialized hardware and software was expensive.
• Poor real-world performance: Demonstrations often did not translate into reliable practical systems.

The effects included reduced research funding, cancellation of projects, declining public confidence, and a shift toward more practical and carefully evaluated AI methods. These periods also encouraged researchers to develop stronger algorithms and realistic performance goals.

An expert system is an AI program designed to imitate the decision-making ability of a human specialist in a particular field.

Its main components are:

• Knowledge base: Contains facts, rules, and domain-specific information.
• Inference engine: Applies rules to known facts to reach conclusions.
• User interface: Allows users to interact with the system.
• Explanation facility: Shows how a conclusion was obtained.

Expert systems became popular during the 1970s and 1980s in areas such as medical diagnosis, mineral exploration, equipment maintenance, and financial decision-making.

Their contribution included demonstrating that valuable expert knowledge could be encoded in computer programs and used consistently. However, they had limitations: knowledge acquisition was difficult, rules were expensive to maintain, and systems usually lacked the ability to learn automatically or handle unfamiliar situations.

Early AI relied mainly on symbolic AI, in which knowledge was explicitly represented using symbols, logical statements, and hand-written rules. This approach was useful for structured problems but had difficulty handling ambiguity, noise, and large-scale real-world data.

Machine learning introduced methods by which systems learn patterns from examples rather than relying entirely on manually written rules. Common approaches include supervised learning, unsupervised learning, and reinforcement learning.

Deep learning uses multilayer neural networks to automatically learn increasingly complex representations from data. Improvements in:

• Computing power,
• Large datasets,
• Specialized hardware such as GPUs, and
• Better training algorithms

made deep learning highly effective for image recognition, speech processing, language tasks, and prediction. Modern AI often combines learning-based techniques with symbolic reasoning and other computational methods.

From 2010 onward, AI experienced rapid growth because of large datasets, cloud computing, improved algorithms, and powerful graphical processing units.

Major developments include:

• Computer vision: Deep neural networks achieved high accuracy in image classification and object detection.
• Speech recognition: AI became capable of transcribing spoken language more accurately.
• Natural language processing: Systems improved in translation, summarization, question answering, and text generation.
• Reinforcement learning: Programs achieved impressive results in complex games and simulated environments.
• Generative AI: Models began producing text, images, audio, video, and computer code.
• Autonomous systems: AI became more important in vehicles, robotics, and industrial automation.
• Responsible AI research: Greater attention was given to bias, fairness, privacy, transparency, safety, and accountability.

Despite these advances, current systems can still make errors, produce unreliable information, and require careful human oversight.

AI is used in healthcare to assist professionals, improve efficiency, and support better patient outcomes.

Applications include:

• Medical image analysis: Detecting possible abnormalities in X-rays, CT scans, MRI scans, and other images.
• Clinical decision support: Providing doctors with information about possible diagnoses or treatments.
• Drug discovery: Predicting molecular properties and identifying promising compounds.
• Personalized medicine: Relating patient characteristics to suitable treatments.
• Patient monitoring: Detecting changes in vital signs and identifying risks.
• Virtual health assistants: Answering common questions and helping patients schedule services.
• Administrative automation: Processing records, billing, and appointment information.

AI does not replace medical professionals completely. Its results must be validated because errors, biased data, privacy concerns, and incorrect recommendations can affect patient safety.

AI has become important in finance and banking because these fields generate large amounts of structured and unstructured data.

Key applications include:

• Fraud detection: Identifying unusual transaction patterns.
• Credit scoring: Estimating the likelihood that a borrower will repay a loan.
• Algorithmic trading: Analyzing market data and executing trades according to programmed strategies.
• Customer service: Using chatbots and virtual assistants to answer questions.
• Risk management: Predicting financial risks and monitoring market conditions.
• Anti-money-laundering systems: Detecting suspicious networks of transactions.
• Personalized financial advice: Recommending products based on customer needs and behavior.

These applications can improve speed and accuracy, but they must be monitored for unfair bias, lack of transparency, privacy violations, and financial errors.

AI supports transportation by helping systems perceive environments, predict events, plan routes, and make decisions.

In autonomous vehicles, AI may perform the following functions:

• Perception: Detecting lanes, pedestrians, vehicles, traffic signs, and obstacles.
• Localization: Determining the vehicle's position using maps, sensors, and satellite information.
• Prediction: Estimating how nearby vehicles and pedestrians may move.
• Planning: Selecting a safe route and driving action.
• Control: Adjusting steering, acceleration, and braking.

AI is also used in traffic prediction, public transportation scheduling, route optimization, logistics, and predictive maintenance. These systems can reduce delays and improve safety, but they must handle uncertain weather, unusual road conditions, sensor failures, cybersecurity threats, and ethical decisions.

In education, AI is used for:

• Adaptive learning systems that adjust content to a student's level.
• Automated feedback and assessment.
• Intelligent tutoring systems.
• Learning analytics that identify students who may need support.
• Translation, accessibility tools, and speech-to-text services.

In e-commerce, AI is used for:

• Product recommendation systems.
• Customer-service chatbots.
• Demand forecasting and inventory management.
• Personalized advertising.
• Search ranking and product classification.
• Fraud detection and payment security.

These applications improve personalization and efficiency. However, organizations must protect student and customer data, avoid discriminatory recommendations, and ensure that automated decisions remain understandable and subject to human review.

Natural language processing, or NLP, enables computers to work with human language in written and spoken form.

AI-based NLP applications include:

• Speech recognition and voice-controlled systems.
• Machine translation between languages.
• Sentiment analysis of opinions and reviews.
• Text summarization and information extraction.
• Question answering and conversational chatbots.
• Grammar correction and predictive text.
• Document classification and search.

Virtual assistants use several AI capabilities together. They receive speech, convert it into text, interpret the user's intention, retrieve or generate a response, and may perform an action such as setting a reminder. Although modern assistants are useful, they can misunderstand accents, context, sarcasm, or ambiguous requests and may produce inaccurate information.

Manufacturing:

• Robots perform repetitive assembly and material-handling tasks.
• Computer vision detects defects in products.
• Predictive maintenance identifies possible equipment failures.
• AI optimizes production schedules and supply chains.

Agriculture:

• Drones and sensors monitor crop health.
• AI identifies pests, diseases, and nutrient deficiencies.
• Smart irrigation supplies water according to field conditions.
• Yield prediction supports planning and resource management.

Cybersecurity:

• AI detects unusual network behavior.
• Systems classify malware and phishing attempts.
• Automated tools prioritize security alerts.
• User behavior analytics can identify compromised accounts.

These applications improve efficiency and early detection, but attackers can also use AI to create more convincing attacks. Therefore, AI systems require continuous monitoring, secure data, and human expertise.

Advantages of AI:

• Performs repetitive tasks quickly and consistently.
• Processes large volumes of data.
• Supports better predictions and decision-making.
• Operates continuously without fatigue.
• Improves personalization of services.
• Assists humans in dangerous or inaccessible environments.
• Can reduce costs and increase productivity.

Limitations of AI:

• Depends heavily on the quality and representativeness of data.
• May reproduce social or historical bias.
• Often lacks common sense and genuine understanding.
• Can be difficult to explain, especially when complex models are used.
• May create privacy and surveillance concerns.
• Can be vulnerable to errors, manipulation, and cyberattacks.
• May affect employment and require workers to develop new skills.

AI should therefore be deployed with testing, transparency, human oversight, security controls, and appropriate ethical standards.