ALP partnered with Virtual Virginia and the Virginia Association of School Superintendents (VASS) to develop beginner, intermediate, and advanced-level AI literacy courses for teachers across Virginia. This blog is the third in a series highlighting content that was created for these courses. Vocabulary like machine learning, large language models, and prompt engineering is prevalent in AI literacy courses. This post explores why this vocabulary is important and how it can help boost your effectiveness when leveraging AI tools.
Machine Learning
Have you ever had to prove that you were human on the internet? You may remember being asked to type distorted letters, check a box labeled “I am not a robot,” or identify every image that contains a stoplight, crosswalk, or bicycle. These tests are called CAPTCHAs. CAPTCHA stands for Completely Automated Public Turing test to tell Computers and Humans Apart. Their original purpose was simple: separate humans from bots. Over time, CAPTCHAs became something else. Each time millions of people labeled images or transcribed text, they were not just proving their humanity—they were training machines to see, read, and interpret the world. Human judgment was used on a massive scale to improve machine learning systems.
CAPTCHA reminds us that AI systems are not intelligent in the human sense. They depend on human-generated data, human-labeled examples, and human-designed rules. What looks like machine “understanding” is actually the result of patterns learned from enormous amounts of human work.
Large Language Models
Generative AI systems are designed to produce new content based on patterns learned from training data. This content may include text, images, audio, video, or code. Large language models (LLMs), such as those behind common chatbots (ChatGPT, Gemini, Claude, etc.), fall into this category.
These systems work by predicting what comes next—word by word, pixel by pixel—based on probability. They do not check sources, evaluate accuracy, or understand meaning. Their strength lies in fluency and pattern replication, not comprehension. Similarly, when a vision model, such as a self-driving car, identifies a stoplight or a pedestrian, it is matching visual features to statistical representations learned during training, not “seeing” in a human sense. Remember the CAPTCHA’s that required you to click on every motorcycle, fire hydrant, or bus?
Prompting
When we write a prompt in a Large Language Model, the AI tool will generate statistically probable content. This is meant to sound fluent and reasonable, not necessarily to share the right answers or invent new solutions to challenges.
Through an educational lens, think about all of the data that Large Language Models may have been trained on. It could be educational textbooks going back decades, Reddit teacher threads, and publicly available teacher blogs online. If you have ever asked AI for a lesson plan and been dissatisfied with the response, it may be because the research on educational best practices in 2026 isn’t as “big” in the AI model as historical educational data. And the historical educational information used to train AI may not be accurate. That’s not to say that there isn’t good educational data in Large Language Models. It just means that when you’re writing a prompt, the machine is using all of the data equally to generate probable outputs.
There are steps you can take to improve your prompting and receive better outputs. When you are prompting for a lesson plan, add specific instructional frameworks that you would like AI to use, add exemplar lessons that you’ve created in the past for AI to model, and add additional resources such as your district’s Portrait of a Graduate. When you add context, resources, and examples to your prompts, you leverage AI to boost your subject matter expertise and narrow the focus to resources that you deem valuable.
Innovative Advances With AI
You may have seen headlines about how AI is transforming certain industries, but you aren’t seeing those changes in education. Let’s use what we know about machine learning and large language models to understand why. As an example, AI has become really good at predicting the weather. In this field, AI models are trained on historical weather data and human-developed weather algorithms. Scientists aren’t asking ChatGPT, Gemini, or Claude to predict the weather. They are building their own AI weather-predicting foundational models with vetted and accurate data sources. This is true across healthcare and technology as well.
While we don’t have the same technology available to us in education, Large Language Models are powerful tools when we learn how to use them based on how they’re made. Perhaps you have a real challenge in your classroom or school that has been persistent for years, and despite your best efforts, the solution eludes you. If you prompt AI to solve a real challenge (i.e, chronic absenteeism), it will pull from public sources and give you solutions that districts are already trying. If you want to leverage AI to actually solve chronic absenteeism, you need to tell it to generate new solutions, think outside the box, disregard current constraints, explore solutions from other industries, etc.
One strategy for finding innovative solutions to educational challenges is reverse prompting. Maybe you have ideas of what you want the AI to generate, but aren’t exactly sure how to word it. Reverse prompting is when you ask AI to write the prompt for you. An example reverse prompt to explore the challenge of chronic absenteeism is: “Help educators solve the challenge of chronic absenteeism. Create a prompt that I can copy and paste into an LLM that will generate never-before-used solutions that think outside the box.” Next, use the prompt that the AI gave you to look for those outside-the-box solutions. You probably won’t get a perfect solution. But you will have creative new ideas that would never exist without creative exploration with AI.
There is a lot of information that is identified as foundational AI literacy for teachers. At ALP, we want to unpack this information in ways that are relevant, practical, and engaging to educators. If you are interested in having your staff go deeper with this content about machine learning, understanding generative and agentic AI, and applying prompting in ways that are ethical, responsible, and effective, reach out! Let ALP help your district equip educators with AI literacy through relevant, competency-based, professional learning courses.
AI Transparency Statement: AI was used to draft parts of this blog with provided research and context. AI-generated text was edited by the author.
References
- “Introduction to reCAPTCHA.” YouTube, uploaded by Google Search Central, 26 January 2010, https://youtu.be/euRAfUGX8wY
- Jones, Nicola. “A.I. Is Quietly Powering a Revolution in Weather Prediction.” Yale Environment 360, 14 Apr. 2025, https://e360.yale.edu/features/artificial-intelligence-weather-forecasting
