Teach with expert insights on AI, curated by your trusty Teacher’s AIde

Welcome back to Teacher's AIed, where we, too, wonder if students just starting to drive could find their way home without their phone’s GPS.

In an age dominated by artificial intelligence (AI), it's not just students who need to grasp the fundamental concepts of how computers perceive the world. Educators, and indeed anyone living in this AI era, can benefit from understanding these principles.

This is the third installment in my series on AI literacy and its implications in K-12 Education.

Did you miss the previous post? Here’s the link:

• What Is AI Literacy? Part 1

• Are you smarter than an AI Literate 5th grader?

In today’s post, I explore AI4K12’s second Big Idea - Representation and Reasoning. While some concepts may seem simple at first glance, they are crucial for comprehending the mechanics of AI.

Picture a map. Even better - picture a hand-drawn map of a neighborhood created by a young student (I remember doing this activity in kindergarten and was honored that my map made it up on our fridge for a time).

What’s the connection to AI literacy? Turns out that creating, reading, and using a map is a great analogy for how AI represents, searches, and reasons with data. We can break this down into three segments:

1. AI tools organize data into a digital representation.

2. AI tools search through that digital representation to find an answer.

3. AI tools, through problem solving, identify the most likely/best answer.

Our post today dives more deeply into these three segment. Let’s jump on in!

1. Representation: Drawing the Map

At the heart of AI is the art of creating data structures that help AI systems understand and interpret information like we do. In simple terms, a data structure is a way of organizing and storing data so that it can be accessed and used efficiently, almost like a bookshelf.

For example, in large language models like ChatGPT, words are often represented as feature vectors - one type of data structure. This numerical representation allows the AI model to process and understand human language. Here’s a helpful image to understand this concept:

What do you notice about the columns in the table? What about the relationship graphed on the right? This is a basic example of how LLMs store information in feature vectors.

This concept of representation is not unique to AI; it's something that humans do all the time. Ever mapped out your neighborhood in your head or sorted objects by their features? If so, then you’ve created a data structure in your own mind!

The key difference is that AI systems use formal, mathematical structures to represent knowledge, while human representation is often more intuitive and less structured. The goal's the same: to arrange info in a way that makes sense and helps us make decisions.

By understanding how computers represent information, educators can help students appreciate the parallels between how computers and humans process information.

2. Search: Looking for Routes

If representing knowledge is like drawing a map, then search, the second concept, is like reading the map to find all the possible routes.

What does this look like with AI?

AI search is all about exploring possibilities to crack a problem.

For example, when an AI system uses a knowledge graph (a representation where nodes represent entities - people, places, things, or ideas - and edges represent the relationships between them), it's essentially navigating a map of interconnected concepts and relationships using an algorithm. By searching through this graph, AI systems can answer questions, solve problems, or even discover new insights.

I think it is interesting to point out that AI and humans approach problem-solving in fundamentally different ways. Computers can systematically explore a vast number of possibilities at incredible speeds, using algorithms to optimize their search. We humans, on the other hand, lean on intuition, past experiences, and a bit of clever shortcutting (heuristics).

To give a personal (and somewhat niche) example, I’m a huge Wordle-er. Even more so, I love looking at the “Wordle Bot” to see my skill, luck, and words left values per guess. Sometimes, the Wordle bot will tell me that I didn’t make a good guess because my guess only eliminated 2 of the 12 possible solutions versus the bot’s guess being able to eliminate 5 possible solutions. Yet, when I read this, I chuckle, “As if I could identify all 12 possible solutions - usually, I’m lucky just to find one potential solution!”

This is search in action - being able to use an algorithm to identify all the possible endpoints.

3. Reasoning: Picking a Route

Lastly, we get to reasoning: using the info we've represented and searched through to draw conclusions or discover something new.

In the context of education, AI algorithms can reason through various types of problems to find answers.

• For example, in classification problems, an AI system might categorize student essays into different proficiency levels based on their features.

• In prediction problems, it could forecast student performance or learning outcomes based on historical data.

• Sequential decision problems involve choosing a series of actions, such as planning a personalized learning path for a student.

• Combinatorial search and heuristic search are used in scheduling and resource allocation, such as optimizing class schedules or assigning students to classrooms.

All of these outputs are the product of AI reasoning.

ChatGPT, our language model pal, specializes in predicting what word comes next. This prediction game is all about reasoning through word probabilities to find the best fit for the conversation.

In conclusion, Representation and Reasoning are key concepts in AI that educators need to understand and teach. By exploring these concepts, educators can help students develop the skills and knowledge necessary to navigate a world shaped by AI.

Stay tuned for more in this series as we continue to unravel the intricacies of AI Literacy and its importance in education and beyond!

Class dismissed!

Lewis Poche & Kourtney Bradshaw-Clay