CPSC 313 58-59 Spring 2022 Artificial Intelligence Randall Lee Narrow AI vs AGI Intelligence - the ability to acquire and apply knowledge and skills. Narrow AI (ANI) – also referred to as Weak AI, goal-oriented AI which is designed to perform singular tasks (eg facial recognition, speech recognition, voice assistants, driving a car, or searching the internet); very intelligent at completing the specific task it is programmed to do. Virtual assistants such as Siri by Apple, Alexa by Amazon, Cortana by Microsoft are examples. Artificial General Intelligence (AGI) - also known as strong or deep AI, a machine with general intelligence that mimics human intelligence & behaviors, learns, and applies its intelligence to solve any problem; can think, understand, and act in any situation that is indistinguishable from humans. AI Architecture Imagine the Main Program of a self-driving car. What would be some of the Controller Subprograms? What would be an App Subprogram within a given Controller? Think about the major elements that enable a car to be driven. Machine Learning Supervised Learning – AI trains on data that has ideal input-output pairings; critiqued and rated by an instructor in order to learn and improve. Semi-supervised Learning - AI learns off of a smaller amount of labeled training data; combines supervised with unsupervised learning. Unsupervised Learning - AI uses machine learning algorithms to analyze and cluster unlabeled datasets; Neural networks, which simulate the neurons of the human brain, are often used in unsupervised learning. Reinforcement Learning - AI will constantly act and perceive, yet only occasionally receives feedback on its behavior. 4 Natural Language Processing & Emotional Analysis Natural Language Understanding - branch of natural language processing that focuses on the ability of an algorithm or process to parse spoken speech and construct a discernable model that facilitates comprehension. Emotional Analysis - considers the inflection of words, voice timbre and pitch, as well as cadence. All these additional data help to frame the context of the user as they are speaking. Search AI & Conversational AI Search AI - process of navigating from a starting state to a goal state by transitioning through intermediate states; AI may ask some questions to more fully understand the context of the question; AI reaches its goal state by determining the most relevant responses. An informed search algorithm uses domain knowledge that AI has learned in some manner over time. Informed search, by the way, is also called Heuristic search. Conversational AI - an extension of Natural Language AI; as AI approaches the goal state of a given query, it will isolate the highest rated or most relevant responses based upon a user’s preferences, topic of discussion, derived meanings, and state of being. AID An Artificial Intelligence Spiritual, Philosophical, and Psychologica ...