Supp. Chapter 4, Lesson 4 Text
Lesson Four: Artificial Intelligence
You may have heard the term artificial intelligence or AI many times in fictional settings. Hollywood movies and science fiction books all love a smart robot or a scary computer program that is bent on world domination. These artificial life forms display human-like intelligence, emotions and adaptability. But what kind of AI can you find in the real world?
Realistic Artificial Intelligence
The kind of human-like artificial intelligence you see in movies and books does not exist. The human mind is the most powerful computer in the known universe. We have the amazing ability to react to our surroundings, adapting and learning with ease. AI programs can sometimes do better than humans in very narrow tasks like playing chess or recognizing faces in a crowd. Modern machines are increasingly powerful and improve AI capabilities every year. But computers are no match for a human's ability to apply common sense and to learn to do many tasks well.
All computers will do exactly what they are programmed to do. The nature of the program may be so complex that it is hard to predict or understand why software behaves the way it does. Complex programs are also more likely to contain bugs that make the behavior even more unexpected. However, despite your Aunt Bethel’s firm belief, there are no gremlins running around in the machine making it do things of its own free will! Nor can machines evolve new capabilities out of thin air, although neural networks can "learn" how to do things based on many attempts to solve a problem.
Example Artificial Intelligence Applications
Our world is becoming increasingly filled with specialized AI routines that handle parts of our daily life. What kinds of AI might you encounter every day? Let's explore a few common examples!
Voice Recognition and Command Processing
If you have used the phrases "Alexa", "Hey Siri" or "OK Google" to make a request, you have used a type of artificial intelligence. Smart-phones and other devices can send your speech to powerful algorithms that convert the spoken words into text. Once the text is understood, additional intelligence is used to understand what you are requesting and complete the task. You can ask for directions, request a song, order a pizza or do many other things that used to require a human-to-human conversation.
This may seem commonplace to you now, but not that long ago, such automated speech processing was virtually impossible. Now, with more powerful computers and advancements in voice-to-text technology, voice recognition has become the standard control interface for a new generation of smart personal assistants. These new devices (like Amazon Alexa and Google Home) can perform Internet searches, set alarms, create new items on your calendar or even remind you of important dates and events.
Desktop and laptop computers are also getting into the voice assistant game. Windows computers now come pre-loaded with an assistant named Cortana and Mac OS X users can use a desktop version of Siri on their computers.
If you have driven or ridden in a car with a GPS system, or used a smart-phone to give you directions, you have experienced artificial intelligence applied to mapping. Given your current location and destination, the mapping AI must plot you the most effective route using a database of known streets. It will then track your progress and give you turn-by-turn directions when necessary.
Advanced mapping programs will even predict traffic patterns and suggest alternate routes to bypass problem areas. For example, Google Maps uses AI to analyze the speed of traffic moving along your route in real-time. The program also incorporates a database of user-reported traffic problems to create a better route to your destination.
Grading and Assessment
Many teachers rely on grading and assessment AI routines to analyze your work. High school and college students may be familiar with services like Turnitin, a popular tool used by instructors to analyze student writing for plagiarism. These tools use algorithms to determine the similarity of two works, based on large databases of digitized sources. If your paper has the same wording as another source, it may be flagged for plagiarism.
Essay grading is another way that AI is making its way into your school life. Reading a large number of essays is very labor intensive. For this reason, some schools are starting to use essay-grading AI programs that "read" essays and judge them based on linguistics, syntax, sentence length and so on.
One of the most prominent users of this technology is the Graduate Record Exam (GRE), commonly used as an entrance exam for graduate schools. The essay portion of this exam is graded using one human reader and one robo-reader. Both readers produce an independent score of an essay. If the scores differ substantially, a second human reader is brought in to settle the discrepancy. By using a combination of human and AI readers, the essays can be graded more effectively than using either one alone.
Have you ever uploaded an image to Facebook, Instagram or another social media service? These social media services can use AI routines to scan the images for faces and automatically highlight or tag your friends. Facial recognition software can be powered by artificial neural networks (algorithms that mimic the structure of the human brain).
Social media platforms continually look for new ways to enhance your online social experience. Consider the following features:
- Facebook may use AI to determine which news items to show you in your news feeds.
- Pinterest may use AI to "see" pins or images that are visually similar to the pins on your own board.
- Snapchat introduced AI facial filters, called Lenses, in 2015. These filters track facial movements, allowing users to add animated effects or digital masks that adjust when their faces moved.
- Recently, both Android and iOS have added similar facial tracking features to their smart-phones.
All of these AI routines process large amounts of specialized data (such as images or a history of your prior interests) to customize output just for you.
If you have watched the news lately, you are aware that many manufacturers are working on self-driving cars. The hope is that a self-driven car is safer and more efficient than one driven by a human. Self-driving vehicles may become a cheaper way to deliver goods and haul freight, because human truck drivers need to eat, sleep and rest.
It is extremely difficult to create artificial intelligence that is good enough to identify and react to every situation that human drivers may face on the road. Our human eyes and brain combine to instantly recognize surrounding vehicles, roads, pedestrians and other unexpected threats. AI drivers must rely on sensors, cameras and GPS systems to understand where they are and what kind of environment surrounds the vehicle. While it may be relatively easy to self-pilot a vehicle on well-marked streets or highways, driving on a dirt road through the woods may present a tougher challenge. What might happen if a car enters a long tunnel where GPS signals are lost?
Some vehicles come with AI "helpers" that don't drive, but provide other assistance for human drivers. For example, an AI routine may warn the driver if they are drifting into another lane. An AI assistant might also apply the brakes automatically when a sudden threat is detected ahead. Human drivers may even let an AI assistant perform simple tasks such as parallel parking. Driving cars is an example where a combination of humans and AIalgorithms, working together, may provide better results than either one of them alone.
Work with Me: Exploring Self-Driving Cars
Are self-driving cars ready for every-day use? In this exercise, you are going to explore some of the technology behind self-driving cars and also think about safety and legal questions that should be answered before they become commonplace on our roads.
How to Complete this Activity
You are going to do your own online research to gather information about the following self-driving topics:
- Sensors - What kind of sensors are used on self-driving vehicles? How might those sensors fail or produce bad data during every-day use?
- Safety Questions - Do self-driving cars provide the opportunity for humans to correct AI mistakes or take control of the vehicle? Why or why not? Find reports where self-driving cars have been involved in accidents, and list the main cause of each accident.
- Legal Questions - If a self-driving car is involved in an accident, who is legally at fault? Is it the human owner of the car, the car manufacturer, the software company that designed the AI system, or some combination of parties? What kinds of laws regarding self-driving vehicles have been created at the state or national level?
- AI Assistants - What kinds of AI "assistant" features are now available on new car models? These AI assistants still require a human driver, but may provide some automated help for certain situations.
Be sure to carefully record and cite your online sources. Self-driving technology and our society's embrace of the self-driving concept changes each year, so there may not be one "right" answer to any question. Be prepared to present your results to the class or to your teacher, if asked.