“But, shift happens,” he said.
Henecker, the director of planning research and evaluation at Freehold, N.J., Regional High School and founder of Ed Tech Paradigms LLC, presented Thursday at Empire State University’s Learning With Innovative Technology conference. He titled his journey as, “Learner, Teach Thyself: Unleashing the Power of AI Chatbots and ChatGPT for Personalized Learning.”
When he taught high school English, Henecker collected writing samples from his students during the first two weeks of instruction. This helped him identify individual writing styles so he could potentially spot plagiarism or lack of source attribution later on. That experience prepared him to work responsibly with generative AI tools such as ChatGPT and Google Bard — as classroom partners where teachers also understand technology’s limitations.
“There are so many detection tools, but they are not perfect — they’re just guesstimates,” he said. “The idea is detect by deduction.”
Henecker used ChatGPT in combination with a variety of other AI tools to submit a proposal for Empire State University’s conference. The proposal was accepted, and he used the same tools to build his slide presentation. Then he showed an online audience of educators what these things can do.
He offered Tome presentation software as an example of a good tool for generating images and slides. For content, he said there are a variety of tools besides ChatGPT, including Perplexity AI, which he demonstrated by typing in a prompt requesting a persuasive-writing lesson plan to explain the concepts of ethos, logos and pathos. An outline with a condensed summary appeared within seconds.
“This would have otherwise taken me an hour to write,” he said. “It’s not going to be perfect. It’s only a start.”
Henecker called Anthropic’s Claude 2 his new favorite tool. In addition to having more mathematical capabilities than ChatGPT, it can upload entire documents and allows for longer prompts than many other AI-enabled ed-tech applications. On the STEM side, it could be used to tutor students on a chapter of statistics. On the humanities side, it can identify examples of irony and foreshadowing in a classic novel.
To maximize productivity and efficiency for all generative AI tools, practice prompting, Henecker said. Teachers are familiar with interactive questioning methods that “prompt” students to ask follow-up questions and facilitate discussion, and they can apply that “why, why, why” way of thinking when interacting with AI.
“It’s all about going deeper,” he said. “If you don’t like what you get [at first], keep prompting with more description.”
There’s even a tool for learning the art of prompting: FlowGPT. It has a top-30 list of prompts for teachers, Henecker noted.
Henecker is training ChatGPT to learn his writing style so he can save time on emails. But he cautioned that all of these machine-learning tools feed human interactions into their ever-expanding data sets. Keep this in mind for protecting student privacy, he said — know what to leave out or disclaim. For example, a user can instruct a generative AI tool not to put their school’s website into its memory.
For both educators and students, Henecker stressed that AI should not be seen as something that should do all of the work, but as a tool to get started, to help the user stay organized, and to create a learning space where the user’s own ideas and creativity can thrive.
“Staring at that blank space is the most horrifying thing in the world,” he said.
