Ed Tip: AI FAQ

Generative AI is a form of artificial intelligence that can create, or generate, new content. These are programs like ChatGPT, Google Bard, Anthropic’s Claude, and Microsoft Bing Chat, which is now called Copilot. With any of these options, you type a question or prompt into a search field—much like you would find in Google—and the chatbot gives you an answer. Among other things, you can ask it to draft emails, write essays, or brainstorm ideas. You can even ask follow-up questions. It feels a lot like having a conversation with the computer.

These generative AI chatbots are called large language models because they are trained on billions of samples of human language use—much of it scraped off the internet. Essentially, these tools are trained to produce outputs based on real human language examples. These large language models basically learn how to put words together into meaningful patterns. They’re not really thinking but rather predicting the next most likely word. That’s also why they can be wrong. They are predicting based on learned examples, and sometimes, they get it wrong. When they do this, we say that they are hallucinating. They’re picking words that might make sense together but are not factually correct. It’s important to understand that these chatbots can, and do, make mistakes. While these tools can be really helpful, they are not a replacement for human thinking. It’s helpful to remember the 80–20 formula, which suggests that even if a chatbot can do 80 percent of the work for us, like drafting an email, the human involved still needs to do the other 20 percent. You need to make sure that the content is accurate and the suggested text actually meets your needs. Think of it as a human writing partner who can be very helpful but is not perfect and can make mistakes—or possibly even mislead you, just to make you happy by giving you an answer.

The best way to get started is to set up an account with one of the major chatbots and try it out. Most people agree that the current leaders are ChatGPT, Google Bard, Anthropic’s Claude, and Microsoft Bing Chat, which is now Copilot. They all work in pretty much the same way. Once you set up a free account, you will be greeted by a text box, where you can type in a question or task for the chatbot to complete. Think of it like communicating with a thought partner. Don’t have it do the work for you; have it do the work with you. A great place to start is asking it to help you brainstorm ideas for a lesson plan. The more details that you include, the better the results will be. Don’t be afraid to ask follow-up questions. Again, think of it like having a conversation with the chatbot. And remember the 80–20 rule. You won’t want to blindly take what the chatbot created and use it without vetting it for accuracy and usability. Spending even 20 minutes with this and trying it out can greatly help to demystify the experience. While there are ways to improve your prompts to get better results through prompt engineering, a simple query will still get you started and provide useful results, especially if you ask follow-up questions or request revised answers.

Yes, and there are more offerings being released all the time. A popular option is MagicSchool. While there is a paid version, you can use quite a few of the features for free. There are dozens of options that help you complete teacher tasks: things like lesson planning, generating text at various reading levels, creating rubrics, designing team-building activities, and more. Technically, you can do all of these things in ChatGPT or one of the other chatbots without going through a tool like MagicSchool. However, what MagicSchool and similar tools do is help improve the quality of results while making the prompt writing process simpler. These specialized platforms are built on top of popular chatbots like ChatGPT. They help refine your prompts by having you fill in specific fields and by asking related questions. Especially if you’re new to generative AI, this can make the experience easier, more streamlined, and more efficient.

To be honest, no one knows for sure. However, most people think that this type of AI has the potential to serve as both a teaching assistant and personalized learning tutor. While these tools won’t replace teachers, they may very well help reduce teacher workload and provide each student with immediate and personalized support. It’s hard to get to every student as soon as they have a question, and AI can potentially help with that. Students can ask the chatbot questions and get personalized coaching as they work through a problem. The best example developed so far may be Khanmigo, a really neat AI tool developed by Khan Academy. It’s still in the developmental stages, but you can sign up to try it out for a small fee, currently $4 per month. It not only helps teachers plan, but it is also programmed to be a student tutor. You can have it do things like co-write a story with you, practice debating a topic, or become a historical figure that you can chat with. Many people who have used it feel that Khanmigo is a sneak peek into where generative AI may be heading.