What teachers need to know about AI in the classroom
By Susan Brooks-Young
If you made a list of education technology trends likely to significantly impact K-12 education in the near future, what would you include? Like most educators, you’d probably mention things such as student data privacy, personalized learning, cyber security, or digital learning environments because these issues receive a lot of public attention. Would you think to include Artificial Intelligence (AI)?
Defining Artificial Intelligence (AI)
What, specifically, is AI? Marvin Minsky, founder of the AI Lab at MIT, said AI is “the science of making machines do things that would require intelligence if done by men” (todayct.us/2DlDeh7). More recently, SAS Institute, Inc., defined it this way:
Artificial intelligence (AI) makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. Most AI examples that you hear about today — from chess-playing computers to self-driving cars — rely heavily on deep learning and natural language processing. Using these technologies, computers can be trained to accomplish specific tasks by processing large amounts of data and recognizing patterns in the data. (todayct.us/2CrG6Hs)
The reference to computers being trained to complete specific tasks by analyzing large amounts of data to identify patterns is an important point for students to understand because it helps them see that the quality of tasks completed is dependent on the accuracy of the data being analyzed.
This definition mentions three subfields of AI that are defined as follows:
Machine Learning (ML): An algorithm that enables a computer to “learn” about something by analyzing large amounts of data and identifying patterns. Example: ML makes it possible for entertainment services to make predictions about new programming a user will like based on that user’s past preferences and those of other users with similar tastes. A classroom example is a program that grades student papers.
Deep Learning (DL): A subset of machine learning, deep learning uses a layered structure of algorithms that mimics neural networks in the human brain. This more complex structure produces AI that appears to be more humanlike. Example: DL makes autonomous cars possible. In education, DL is used to support personalized instruction.
Natural Language Processing (NLP): Facilitates human/computer communication based on use of speech and text. Example: NLP makes personal digital assistants like Alexa, Cortana, and Google Assistant possible. An education application is the use of NLP to help students learn a new language.
There are additional subfields related to AI, but these provide a place to get started.
Who is already using AI?
Interest in AI and its potential impact on society in general, including education, is growing for a variety of reasons. For example, in mid-2017, the Institute for the Future’s report, “The Next Era of Human-Machine Partnerships” forecast that artificial intelligence is one of several emerging technologies that will transform the workplace by 2030. Then, in mid-2018, Technavio’s “Artificial Intelligence Market in the US Education Sector 2018-2022” report predicted a Compound Annual Growth Rate (CAGR) of nearly 48 percent in use of AI in K-12 education.
During the 2018 holiday season, Amazon reported that Alexa devices were top sellers, not just in the United States, but in countries around the world where Alexa devices are available. In fact, the devices were so popular in Europe that for a period of time on Christmas Day, the service crashed when too many new owners attempted to set up their devices at the same time.
AI was a hot topic at the Consumer Electronics Show (CES) in Las Vegas, which focused on how AI relates to business and, in turn, our personal lives. Multiple sessions were offered on speech recognition, AI that enables users to interact with technologies just by talking out loud. In today’s world, that may mean interacting with a digital assistant like Alexa or Google Assistant or perhaps a smartphone, but how will voice interfaces impact devices in the future?
At CES, there were also sessions about how AI is impacting the field of biometrics, the use of someone’s unique physical characteristics for purposes of identification (think iris scans or thumbprints), as well as AI applications in health, cybersecurity, transportation, and other fields that touch our lives in some way, shape, or form. LG Electronics announced several AI enhancements to its home entertainment devices and appliances. Samsung is incorporating Bixby, its AI platform, into televisions, smart appliances — such as refrigerators and washing machines — as well as other devices it manufactures. BMW and Honda are working on autonomous (self-riding) motorcycles, and Daimler revealed its self-driving truck.
How does this apply to K-12 education?
Many people still think of AI as a futuristic but not yet feasible reality. However, the truth is that we rely on AI every day without thinking about it. For example, do you use Siri, Cortana, or Google Assistant on your smartphone or other device? Each of these tools is an AI-powered personal digital assistant that responds to voice commands or questions. What about apps such as Google Maps or Waze that offer navigation assistance when you are driving? Both use machine learning (ML), a type of AI, to help you get to your destination.
When you log into a Netflix or Barnes & Noble account, do you see a list of movies and TV shows or books customized to your viewing and reading tastes? These recommendations are generated based on an AI model that relies on your past preferences to predict what you might want to view or read next. Have you chatted online with a customer service “assistant”? Odds are you interacted with a chatbot, not a human being.
AI applications are also making their way into schools and classrooms. The tools currently in use are fairly basic. They include machine-learning-based teaching assistants that customize skills practice for individual student needs and AI applications that can grade multiple-choice tests. Educators familiar with the SAMR Model, which identifies ways technology use may impact teaching and learning, will realize that uses of this type fall in the Substitution and Augmentation levels of the model. Primarily, the AI applications in use at this point are employed to automate tasks that would have been done anyway or to provide feedback more quickly than might have been possible before. This is a natural start to use of AI in classrooms, but it’s important that we not stop there.
This is just one of the reasons it’s essential for educators to start learning about AI now. By engaging early on, we can help direct the course that AI takes in education. In addition, and perhaps even more importantly, it’s critical that we help our students understand what AI is and how it works so they realize that it is yet another tool that can be used to improve their lives, but that it is just that — a tool.
What do we need to teach students about AI?
Ultimately, students should be offered opportunities to explore the computer science behind AI, but initially they need to understand some basics about AI that most adults are still learning. Even very young children can grasp these ideas at a rudimentary level, which can then be built on over time.
Become AI-aware. Stop and think about the different ways people interact with AI every day. As mentioned above, personal digital assistants, navigation systems, and online shopping accounts are just a few of the ways AI has quietly crept into daily life. Some of these AI tools are obviously computer-based, but others may not be so readily identifiable. For example, unless adults and older students have read or heard that online customer service assistants are often chatbots, they may not realize that it is possible to resolve a basic problem online without actually interacting with a human being. This is because many online shopping sites are not transparent about their use of chatbots to streamline customer service.
When adults and older students use social media platforms such as Facebook Messenger to order food or find a movie, they may
understand that they are dealing with a bot to complete the transaction, but what about other social media bots that follow human beings on social media for the purpose of spamming their accounts with inaccurate information? In part, users are fooled because the person behind the bots tries hard to make them appear to be human with plausible usernames and profile pictures.
This blurring of lines becomes more difficult with younger students. Robotic devices brought into classrooms are often anthropomorphized by the adults in the room. The devices are given human names, referred to as “she” or “he,” and ascribed other human characteristics. This can become even more confusing when the device responds to verbal commands through NLP. Very young children are able to distinguish between living and non-living things, but it’s still important that adults reinforce that differentiation: Regularly remind students that AI devices are not human or capable of human judgment or emotions.
Help your students become more aware of the AI around them by talking about it. Brainstorm examples of AI in their daily lives and talk about the kinds of tasks these AI devices are able to do. Ask students what they think about ascribing human characteristics to bots, robots, or personal digital assistants. Why do people do this, and does it matter?
It’s all about data quality. Using ML, computers are trained to perform specific tasks by analyzing large amounts of data to identify patterns that enable them to complete the defined task. For example, with Google’s free AI tool, Quick, Draw!, users draw an object on the screen, and the program “guesses” what it is. Each drawing is added to a database used to train the program to identify what the sketch represents. Teachers often use this tool with students of all ages as a way to introduce the concept of ML.
One second-grade teacher I spoke with used this tool with her students. She asked them each to draw a cat using Quick, Draw! Then she asked students to think about common elements in their drawings of cats, such as a circular head with triangle ears, lines for whiskers, an oval body, and a tail. She wondered aloud what would happen if the students drew a cat using just squares and rectangles. Would Quick, Draw! still recognize the sketch as being a cat? The children discovered that these new sketches were not recognized as cats — but could be if enough of these drawings were added to the database. This led to conversations about how much bad data it takes to corrupt the ML process. Pretty sophisticated thinking for seven-year-olds!
Another important piece of learning about ML is that even when a computer “learns” how to complete a task, that training is very narrow. This is true even with DL. As long as training conditions are met, the task can be completed, but as soon as a variable is altered, computers are not yet able to generalize their learning to the point that they can still complete the task.
There is much more to learn about and consider when it comes to AI and its potential impact on society and education. While there is potential for misuse, the possibilities for improving people’s lives are great.
A former Catholic school teacher, Susan Brooks-Young spent 23 years as a teacher and administrator. She now works as a professional consultant and author. She is co-author of Pathways to Well-Being: Helping Educators (and Others) Find Balance in a Connected World (ISTE, 2019).
All content copyright © Today’s Catholic Teacher/Bayard.com. All rights reserved. May be reproduced for classroom/parish use with full attribution as long as the content is unaltered from its original form. To request permission to reprint online, email firstname.lastname@example.org.