ARR: Lesson - Making Decisions

A machine, taking on a human form, strikes a thoughtful pose.Making Decisions

Think About It...

As we begin our first lesson, ask yourself a few questions:

  • How do you make decisions?
  • What weighs most heavily for you when making a decision? How many factors do you consider?
  • Do you think you could translate how you think to how a machine thinks?

Reviewing Machine Learning

In a previous module, we discussed how machine learning works. In that lesson, we discovered that machines make decisions based on data. Before we move on, let's review the types of decisions that machines make based on data.

  • Classification: A classifier is like a detective that examines something and decides what category or group it belongs to. These categories or groups are called classes.
  • Prediction: A predictor is like a solver that tries to figure out a number. If the predictions are about the present it is called an estimate. If the prediction is about the future it is called a forecast.
  • Recommendation: Recommender systems are like helpers that suggest things to you from a big collection, using what they know about what you like and what people with similar preferences to you also like. Recommender systems base their decisions on things such as:
    • Things you enjoyed before
    • People who like the same things as you
    • Things those people liked that you haven’t tried yet
    • Based on these preferences, the recommender system suggests those things to you.
  • Planning: Planning systems schedule problems which helps in determining smart ways to complete tough tasks in time sensitive situations

AI Decision Making

It is important to note that AI is capable of multi-tasking at speeds that humans are unable to replicate. Meaning that it does not choose one single action to employ at a time but can employ multiple tasks simultaneously…in seconds. In addition to classifying, predicting, recommending, and planning, AI utilizes reasoning models such as decision trees and search trees using breadth-first search (i.e. route finding).

Let’s explore decision trees. A decision tree is a tool that helps make decisions by asking questions and following different paths based on the answers. Each question leads to another question until a final decision is made. Many of the choices that we as humans make on a daily basis could be illustrated in a decision tree. For example, choosing what to wear daily, choosing what to eat/drink, and choosing to participate in certain activities.

Here is a simple example – What shoes should I wear?

Will it rain today?
Yes No
Wear rain boots Wear sandals

So, our decision tree is simple: if it is raining (yes), we wear the boots, but if it is not raining (no), we wear sandals. Can you picture a far more complicated decision tree?

Video Lesson

To reinforce learning and see a visualization of AI categorizations and decision trees, watch the first five minutes of the video below. You may watch the full video, but the key lesson is contained in that first five minutes.

Video Source: Crash Course (YouTube)

[CC BY-NC-SA 4.0 Links to an external site.] UNLESS OTHERWISE NOTED | IMAGES: LICENSED AND USED ACCORDING TO TERMS OF SUBSCRIPTION - INTENDED ONLY FOR USE WITHIN LESSON.
Shutterstock AI Generator/Shutterstock.com. Image used under license from Shutterstock.com and may not be repurposed.