In 2017, Google made what’s called a “pre-trained transformer,” which would later stand for the “PT” in ChatGPT. This “robot,” so to speak, was trained on about 45 terabytes of information, the equivalent of 500,000 lifetimes. Using that information, it then learned from itself, meaning it could generate content, but we couldn’t communicate with the AI. That’s when OpenAI stepped in with the idea of combining it with its natural language processing model which would allow us to communicate through text with the transformer, and the rest was history.
“This natural language processing allows us to communicate with all of this data,” said Holly Clark, owner of The Infused Classroom during the first Future of Education Technology webinar on Wednesday. “It’s not just the data, it’s these neural networks happening in the AI that’s looking at the data for information and then coming back with its own response. No one is telling it to say this.”
To put it simply, the 45 terabytes of data make up the large language model or LLM.
“In my personal opinion, I think that every educator and every human needs to understand this large language model,” said Clark.
The webinar, titled “How AI Will Transform K-12 Education,” explored how AI will inevitably transform how educators curate and deliver instructional content.
“It really does unlock lots of opportunities for us,” said Matt Miller, author of Ditch That Textbook. “It creates things for us.”
For instance, in a matter of six minutes, Miller once detailed more than 10 ways ChatGPT can be utilized in the classroom to promote effective, seamless instruction.
As an example, he put himself in the shoes of a history teacher preparing to discuss the Peloponnesian War. This is how he said he asked ChatGPT for help.
“I asked it all of these questions like, ‘Can you write me a five-paragraph essay about the Peloponnesian War?’ It does it. ‘Can you write it at an 8th-grade level?’ It does it. ‘Can you write it at a college master’s level?’ It does, and of course, the language gets more sophisticated. ‘Can you summarize bullet points for it that I can use for PowerPoint slides?’ And it writes a whole bunch of text I can put on PowerPoint slides.”
“If you’re wondering, ‘How is this going to impact me?’ You might go watch that video and it starts to open your mind to what it can potentially do to help you out in the classroom,” said Miller.
But AI has been around for decades, argued Education Strategist and Consultant Ken Shelton. The technology itself, however, is simply evolving.
“For those of us of a particular age demographic, I remember when I was using Microsoft Word and I made a punctuation error and Clippy would pop up and tell me that I need to add a period or something,” he said. “That was AI. It’s been around for decades. It’s just that you’re seeing the evolution of it.”
While the nature of AI technology has only gotten more impressive over the years, Shelton believes that just because you can automate something doesn’t mean you should. Hear how humans play a significant role in creating biases in generative AI tools in this short clip:
Another point of concern shared by many educators when ChatGPT first launched surrounded that of student plagiarism, an idea that each of these experts argues we must reimagine now that AI has entered the conversation.
Clark, the author of The AI Infused Classroom, outlines in the first chapter of her book how ChatGPT helped her streamline the writing and publication process. She started by writing the draft in a Google document, reviewing and making all the necessary changes.
“At this point in 2018, I would’ve sent it to an editor where it would’ve stayed for six weeks. She would’ve sent it back to me with some recommendations. I would’ve made those recommendations. She’d look at it again then it would go back for grammar and all kinds of stuff. And the whole process would’ve taken between three to six months.”
After making her revisions in Google Docs, she copy and pasted the draft into ChatGPT and asked it to give her some recommendations.
“It would give me five strengths and five weaknesses,” she said. “I would look at those and automatically make those changes. Then, because I was a good prompt engineer, I asked it, ‘Are there any holes in this chapter that the reader might want to know more about?’ And ChatGPT would offer some things and say, ‘Yeah, you might want to talk about this.'”
Once it was edited exactly how she wanted it, she asked the AI to check it for grammar. In nearly half an hour, she said she did what would normally take six months of back-and-forth with an editor.
“What we need to think about is the beauty of the editing process that happens with this AI,” she said. “The beauty of how quickly I can get a book out to you now so that we have something relevant that hasn’t been on a shelf or being edited or talked about for two years before you get that information.”
“Would you call that cheating? Did I cheat in any way, shape or form? Did ChatGPT write my book? Absolutely not! Did ChatGPT help me get it to you sooner? Yes. And that is where the beauty lies.”
And that’s exactly the framework educators need to adopt for their students. How can leaders help curate prompt engineers among their students so that they—and their teachers—can create more personalized, streamlined educational instruction? This is the beginning of the education of the future.
Watch the entire webinar on-demand here.