Intro to Artificial Intelligence

Lecture 1: Introduction to Artificial Intelligence

• Dr. Doaa El-Shahat is the lecturer for CS203 Artificial Intelligence at Zagazig University, Egypt.
• The topics include the definition of intelligence, AI in daily life, the definition of AI, the history of AI, and the foundations of AI.

AI in Daily Life

• Digital assistants like Siri (Apple), Alexa (Amazon), and Cortana (Microsoft) are examples of AI.
• Google, Opera, Firefox, Safari, and Internet Explorer are AI-powered search engines.
• Social media platforms like Facebook, Instagram, YouTube, and TikTok use AI.
• AI is used in translation (Google Translate), writing assistance (QuillBot, Grammarly), chatbots (ChatGPT), and search (Deepseek).
• AI is present in iRobot Roomba vacuum cleaners and self-driving cars.

Definition of Intelligence

• Intelligence is the ability to learn, solve problems, and use knowledge to adapt to new situations.

Definition of AI

• AI definitions fit into four categories: thinking humanly, thinking rationally, acting humanly, and acting rationally.
• Thinking humanly involves cognitive modeling.
• Thinking rationally involves using logic through "laws of thought".
• Acting humanly involves passing the Turing test.
• Acting rationally involves the "rational agent approach."
• Thinking humanly emphasizes cognitive modeling.
• Thinking rationally uses laws of thought approach using logic.
• Acting humanly seeks to pass the Turing test.
• Acting rationally aims for the rational agent approach.
• Acting Humanly can be defined as the art of creating machines that perform functions that require human intelligence
• Cognitive modeling approach seeks to create Machines with Minds

Approach 1: Acting Humanly

• The Turing test is an ultimate test for acting humanly, testing machine intelligence.
• A computer and human are interrogated by a judge; the computer passes if the judge can't tell the difference.
• Alan Turing (1950) was an English mathematician and pioneer of computer science.
• For a machine to pass the Turing Test, it needs:
  • Natural language processing to communicate in English.
  • Knowledge representation to store information.
  • Automated reasoning to answer questions and draw conclusions.
  • Machine learning to adapt and extrapolate.
• Notable AI machines include ELIZA (1966), PARRY (1972), ALICE, and Eugene Goostman.
• Sophia and BabyX are other examples of AI attempting to act humanly.
• Intelligent systems are designed to act like humans because humans are often the gold standard for many tasks.
• Humans needed for innovation, critical decision-making, and fields requiring emotional intelligence.
• A Total Turing test includes physical aspects, involving computer vision and robotics.

Computer Vision

• Computer vision uses computer science to allow computers to see and understand digital images or videos.
• It uses object detection, semantic segmentation and classification/recognition.
• Semantic segmentations are where the pixels belong to the object in the image

Robotics

• Robotics is a branch of engineering and computer science for the conception, design, manufacture, and operation of robots.

Natural Language Processing (NLP)

• NLP is a machine-learning technology for computers to interpret, manipulate, and comprehend human language.
• NLP uses include summarization, speech-to-text conversion, sentiment analysis, text classification, spell checking, grammar checking, question answering, and voice assistants.

Approach 2: Thinking Humanly

• Knowing if a program thinks like a human requires understanding the human mind.
  • Introspection (catching our own thoughts).
  • Psychological experiments (observing behavior).
  • Brain imaging (observing brain activity).
• Once a theory of the mind is sufficiently precise, it can be implemented as a computer program.

Approach 3: Thinking Rationally

• Systems yield correct conclusions when given correct premises.
• Expressing informal knowledge in logical notation is challenging, especially if the knowledge is not 100% certain.
• Also a big difference exists between a problem "in principle" and solving it in practice.

Approach 4: Thinking Rationally

• Study intelligent agents.
• Also known as the the "the rational agent approach”
• It is the computations that make it possible to perceive, reason, and act

History of AI

• 1950: The "Turing Test" was developed.
• 1956: The term "AI" was introduced, and the first AI software program was created.
• 1967: The Mark 1 Perceptron, a computer that "learned" through trial and error.
• 1968: The Perceptrons Book became a key work on neural networks.
• 1980: AI becomes widely used in applications and is widely used since then.
• 1995: the "Artificial Intelligence: A Modern Approach" textbook was published
• 2004: the question of what is artificial intelligence was researched
• 2015: human image recognition accuracy occurred
• 2016: Google then acquired DeepMind for $400 million.
• 2023: AI was boosted with generative tech and big data.
• Timeline:
  • AI evolved from Weizenbaum's chatbot (1964) to virtual assistants like Siri (2011) and Alexa (2014).
  • Chess-playing computers beating grandmasters (1997) and automated vacuum cleaners (2002).

Foundations of AI:

• Philosophy explores formal rules and the connection between knowlegde and action.
• Mathematics provides formal rules to draw conclusions.
• Neuroscience studies how the brain processes information.
• Psychology examines how humans and animals think and act.
• Economics centers on maximizing profit.
• Computer engineering focuses on building efficient computers.
• Linguistics analyzes how language is related to thought.