Artificial Intelligence Overview

Study Notes

In 2022, a Google engineer was fired for claiming a chatbot he was working on had become sentient, perceiving and expressing thoughts and feelings equivalent to a 7- to 8-year-old child.
This event highlighted concerns about the capability and secrecy surrounding artificial intelligence.
The program General Problem Solver, developed in 1956, aimed to mimic human problem-solving using Means-End Analysis.
While successful with logical problems, it struggled with real-world applications.
A scenario similar to John Searle's Chinese Room experiment illustrates that a system can produce correct outputs without understanding the underlying language.
An activity involves forming teams to define "strong AI" and "weak AI".
Arthur Samuel, in 1959, developed a game capable of learning strategies via trial and error, defining Machine Learning.
Machine learning involves computers deducing logic to complete tasks without explicit instructions.
Supervised learning trains AI with extensive inputs and outputs to enable autonomous task completion.
Unsupervised learning provides AI with inputs without outputs, enabling it to identify patterns and relationships.
Reinforcement learning differentiates from conventional input-output paradigms, focusing on an agent interacting and adapting to an environment to achieve optimal actions.

AI has revolutionized healthcare through diagnostics, personalized medicine, drug discovery, and robotic surgeries.
In finance, AI aids marketing, trading, fraud detection, customer service, and automating business processes.
AI is a key driver in transportation advancements, including autonomous vehicles (self-driving cars and drones), enhancing transportation systems, and optimizing traffic flow.
The retail and e-commerce sectors benefit from AI through personalized recommendations, efficient inventory management, and chatbot customer service.
AI integrates with Internet of Things (IoT) devices in smart homes to enhance services like controlling lights, security cameras, and more.

Generative AI distinguishes itself by creating novel content (text, images, music, etc.) based on learning patterns from vast datasets.
Generative AI's capabilities include generating images, text, and more, drawing from prior input and training.
Generative AI has applications in diverse fields like art, entertainment, design, literature, and scientific research. It can create lifelike images, compose poems or stories, or develop models for investigations.
Generative AI's variability and unpredictability in outcomes present some ethical considerations, particularly concerning originality, copyright, and responsible use.
The types of generative AI include text-to-text, text-to-image, image-to-image, text-to-video, and text-to-voice generation.