AI/Robotics Extension Quiz
What is artificial intelligence?
AI is software that enables an IS to mimic or simulate human intelligence 3 AI techniques: 1. Expert Systems are rule-based systems that encode data from human knowledge in the form of If/Then rules. Such rules are statements that specify that if a particular condition exists, then some action should be taken. One example is the rules that could be part of a medical expert system for diagnosing heart disease. An inference engine applies logical rules to the knowledge base to derive new knowledge. 2. Content management systems (CMS) support the management and delivery of documents including reports, Web pages, and other expressions of employee knowledge. Typical users of content management systems are companies that sell complicated products and want to share their knowledge of those products with employees and customers. Challenges: First, most content databases are huge; some have thousands of individual documents, pages, and graphics. Second, CMS content is dynamic. Documents do not exist in isolation from each other. Documents refer to one another, and when one changes, others must change as well. A fourth complication is that document contents are perishable. Documents become obsolete and need to be altered, removed, or replaced. Content is provided in many languages 3. A neural network is a technique of AI inspired by the networks of neurons in our brains and central nervous systems. A neural net is an adaptive system of nodes and connections that learn. Much like the collection of neurons in our bodies, each node in a neural network connects to many other nodes, and working together the network of nodes produces an output. This neural net compares the output to a right answer, and the network adjusts how the neurons fire and influence each other in order to produce a better output the next time. 5 components of AI IS 1. Hardware- Traditional computing system or a robotic device 2. Software- Unconventional algorithms that learn, are goal centered, and are narrow and specific 3.Data- Domain specific, often rules or statements (e.g., "faces are round," "dogs are animals") 4.Procedures- Methods for converting implicit expertise or the steps to get expertise 5. People- Experts and end users narrow AI: using AI to accomplish a narrowly defined intelligent task such as generating a recommendation, spotting a fraudulent financial transaction, understanding a sentence, or driving a car. On the other hand, broad AI is a general intelligence that can be applied across a broad set of tasks such as counseling, planning, or goal setting Another example of broad AI is intelligent conversation, and a famous conversational challenge for AI is the Turing test. An AI passes the Turing test if it can fool individuals into thinking they are talking with a real person.
How are robots used in business?
AI supports processes, and robots are physical machines that do work, often replacing people. Attributes of a Robot: Autonomy means the ability to operate, at least in part, without direct human intervention. Most robots are also lifelike; many of them perform an action like humans or animals. Finally, all robots move and create action. Robots typically perform roles that are dirty, dangerous, dull, or distant. These four D's are a good way to determine if a role might be more suitable for a robot than a person. Challenges of Robots: Human Unemployment, Harm to human life, Vulnerable to threats Strengths of Robots/AI: Productivity, Speed, Quality, Scale, Consistency Few Injuries Strengths of humans: Low development cost, better security, Flexibility, Judge Ambiguity
What are the challenges of AI?
Expense: AI is difficult and expensive to develop. A single AI application can require many labor hours from expensive experts in the domain under study and AI designers High Expectations: Proponents of AI hoped to be able to duplicate the performance of highly trained experts, like doctors. It turned out, however, that no AI has the same diagnostic ability as knowledgeable, skilled, and experienced doctors. Even when AI came close in ability, changes in medical technology required constant changing of the AI, and the problems caused by such changes are very expensive to correct Limited Understanding of Human Intelligence: while AI can mimic the human mind, our scientific understanding of the mind is still limited. As a result, progress in AI, even narrow AI, is limited
What is the intelligence process?
Intelligence- the ability to acquire, store, and apply knowledge Intelligence Process activities: acquire, store, and apply knowledge This is a dynamic process, as the activities do not always follow the same order. For example, many applications can occur after data is stored, and new acquisitions of knowledge can be added after storing has occurred. The goal of these activities, of the intelligence process, is effective behavior by the user or by the AI. Knowledge is justified beliefs. For example, I know a stop sign means I must stop my vehicle—my belief that I should stop is justified by the law Knowledge management vs Intelligence process Knowledge management also acquires, stores, and applies knowledge, but it broadens the Acquire activity to include creation
What business processes are supported by AI?
Medical Diagnosis: Patients can use AI expert systems to diagnose and better understand their own illnesses. One expert system, the Web site WebMD, asks the patient a series of questions about their ailment, runs their answers through its expert system, and produces a diagnosis. Locating Expertise: Microsoft uses SharePoint to help its coders write software. Coders frequently update their SharePoint profiles with software applications they helped develop. For example, Cal Smith helped write the code for Word tables in Word 2013. If during the development of Word 2020 a coder at Microsoft runs into a problem with the autowidth function of tables, she can use Share-Point to find and communicate with Cal. Fraud Detection: Credit card companies use neural network AI to identify fraudulent transactions. Nodes in the neural net include "has this card been used more than three times in the last 15 minutes," "is this transaction near where the cardholder resides," "is this transaction online," and "is this transac-tion for a high-risk item." The output from the neural network is a yes/no decision, and the network is provided the feedback on its decision so it can update its weights. Human Authentication: Facebook built a neural network to recognize and authenticate human faces. This program, called DeepFace, can accurately differentiate human faces about 97 percent of the time.
