In this lesson, you’ll continue exploring how AI tools work and the role of AI models trained on data. You’ll learn more about generative AI for creating new content and AI tools that understand natural language. Then, you’ll experience an AI tool firsthand as you consider creative ways to apply it in your practice. Get ready to unlock the potential of AI for education!
Generative AI and its applications
AI tools can appear seamless and intuitive, but behind the scenes are powerful models at work. To better understand the relationship between AI tools and their models, take a peek under the hood. Picture AI tools as the user-friendly dashboard of a car. While the dashboard displays information and allows you to give basic commands, the car’s true power comes from its complex engine and intricate network of parts. Similarly, AI models are the hidden engines that power these tools, performing the complex calculations that underlie their functionality. In other words, the AI tool is the interface, enabling you to provide prompts and instructions. This information is then relayed to the powerful AI model (the engine), which does the heavy lifting and generates the response.
When developing AI models, the training process involves six steps:
-
Define the problem: AI designers have a goal in mind, consider AI’s capabilities and limitations, and identify a solution.
-
Collect the data: Designers collect relevant data with which to train the model.
-
Prepare the data: The data is prepared for training. This includes filling in missing data or removing biased data that might mislead a model.
-
Train the model: The model learns patterns in the training dataset and makes successful predictions over time.
-
Evaluate the model: Performance is assessed to ensure accurate predictions will be made outside of the training environment.
-
Deploy the model: Finally, once the model is performing well, it’s deployed within an AI tool.
Training an AI model is an iterative process that involves repeated adjustments until a reliable tool is achieved.
AI tools
Now that you understand more about how models make AI tools possible, it’s time to delve into AI tools. You’ve learned that conversational AI is a type of AI that can understand human language requests and generate responses in a meaningful way. Conversational AI tools are useful because they understand natural language — the way people talk or write when communicating with each other. This means they can process human language requests and generate responses in a meaningful way, respond to common conversational cues, interpret certain colloquialisms, adapt to particular conversational contexts, and engage in natural conversations. However, note that AI may encounter problems when trying to understand more complex aspects of human language, such as emotions, cultural references and nuances, and sarcasm and irony.
Many conversational AI tools are based on large language models (LLMs). These are AI models trained on large amounts of text, which enables them to identify patterns between words, concepts, and phrases in order to generate effective responses to prompts. LLMs work by predicting the next-most-likely word or words in a sequence, using context clues.
If you type in a prompt field, “Debate the pros and cons of …”
Depending on what the LLM has learned from your search history, some predictions could be …
- cat versus dogs
- tennis versus pickleball
- TV shows versus movies
Or perhaps you type, “The difference between …”
In this case, the LLM could predict that you want to learn about the difference between …
- mean and median
- mitosis and meiosis
- affect and effect
Understanding how LLMs predict word sequences will help you navigate conversational AI tools more effectively because you can adjust your prompts to get the most relevant responses.
Begin using AI
Now that you’ve learned about LLMs, it’s time to witness one in action. Here’s an example using Gemini. This prompt asks for a simple explanation of genAI for middle school teachers that includes relatable music analogies. Then, Gemini generates an output fulfilling the request. Check out the response:
As this output demonstrates, using a conversational AI tool is truly like having a conversation!
Disclaimer: Results for illustrative purposes.
Now you try it
It’s time to start your own conversation! As you do, you’ll follow these steps. Feel free to copy and paste them into a document or save them another way so you can refer to them anytime:
Step 1: Access an AI platform
You can use browser-based tools, such as Gemini or ChatGPT. This activity uses Gemini.
To access Gemini:
-
Go to gemini.google.com.
-
Sign in to your personal Google account.
Refer to the resource about how to Create a Google Account, if you don’t already have one. For further assistance signing into Gemini, please refer to Gemini Apps Help.
Note: Before you use Gemini, review the following information:
-
You must be over 18 years old to use Gemini.
-
Review the Gemini Apps Privacy Notice.
-
Please don’t enter private or confidential information in your Gemini conversations or any data you wouldn’t want Google to use to improve its products, services, and machine learning technologies.
-
Gemini is not available in certain countries and languages. For more details, refer to documentation about Where you can use the Gemini web app.
-
Feedback from a wide range of experts and users helps Gemini improve every day. So when you try Gemini, you can provide feedback using the thumbs up or thumbs down button—with the option to further explain in a comment.
Step 2: Locate the prompt field
The prompt field is usually a text box or bar labeled “prompt,” “query,” “input,” or something similar. Imagine it as the opening line of your conversation.
Step 3: Type the prompt
Type your prompt clearly, being specific about what you want the AI tool to do, answer, or create. The clearer your prompt, the better the AI tool can understand and respond. For this activity, input:
Prompt:
| Write a short, happy poem in the style of a wise sorceress. |
Step 4: Press enter
Press enter or submit. This initiates your chat! The AI tool will process your prompt and generate a response. This might take the form of text, code, images, and lots more, depending on the tool and your unique request.
Step 5: Review the output
Your output will be different from what you see here because every AI tool is different — that’s OK! Did you like your poem? If so, great! If not, think of a way to refine your input to improve the output. For instance, you could ask the AI tool to write the poem about a specific subject, reduce or lengthen the word count, or write it in the style of a different persona.
Output:
Use AI to help you prepare for a trip
Continue your practice and prompt an AI tool to help you prepare for a trip. For example, you could type:
Prompt:
| I am a parent. Generate a packing list for a two-night summer camping trip at a lakeshore with four adults and three kids. |
Type your own prompt now, then review the output.
Note: The output displayed was provided by Gemini. Your output may differ from what’s displayed.
Output:
Again, feel free to keep the conversation going until you’ve planned the perfect getaway! You can change up your favorite vacation activity, destination, and much more.
As you continue using AI tools for education, there are a few things to keep in mind: First, it’s possible for these tools to arrive at incorrect answers to math problems, to attribute quotes or work to the wrong person, and to provide erroneous information — especially about recent events, as there’s not enough training data available yet. Likewise, different AI tools have different functionalities. Therefore, they will respond to prompts in their own unique ways. As these tools improve, they should produce stronger results over time.
Continue practicing writing and refining your instructions in order to translate them into creative and helpful outputs. Remember, there’s no substitute for the human touch; so always use your expert eye to ensure the results are effective and appropriate.