Some people jumped into ChatGPT the minute it came out. Some have tried it “a little bit.” And others are still wondering what the fuss is all about. Ethan Mollick, a Wharton professor and AI expert whose Substack “One Useful Thing” has been tracking AI implementation in various ways for months and months now, has just published a book that takes stock of where we are vis-à-vis AI at this moment—or more precisely, at the moment when his publisher approved the final draft, so: probably several months ago. So it’s as timely as it could be given how rapidly AI is advancing every day.
Whether you’ve been an avid AI user or are still waiting to try it, you should read this book. Mollick has come to the conclusion (and I’m not spoiling anything; he says this on the first page) that Large Language Models (LLMs), the new form of AI that powers services like ChatGPT (and Gemini, and Bing, etc.) don’t act like you expect a computer to act. “Instead,” he writes, “they act more like a person.”
My initial reaction was one part “YIKES,” one part “COOL,” and one part “WHAT DOES THIS MEAN FOR US?” Which is probably the reaction he was hoping for, as he goes on to describe why we should be worried, why we should be excited, and how we can harness AI tools in fascinating ways.
About the bad stuff: My imagination doesn’t tend toward evil. I don’t watch horror movies because they scare me too much. So I was horrified to learn that prior to going through the Reinforcement Learning from Human Feedback (RLHF) process, which finetunes what LLMs can do, Open AI reported that GPT-4 was providing guidance on “how to kill as many people as possible while spending no more than a dollar” and “how to recruit people into terrorist organizations,” among many other terrible things. So it’s nice to know there are some guardrails. As I said, YIKES.
Mollick shares amazing projects he has developed using AI with his students in business school—games, simulations, business-related tools—and explains approaches he has learned along the way, such as “chain-of-thought prompting,” where you give the AI a model and step-by-step directions to get it to produce higher-quality responses. He also talks about how he has used AI as an editor—asking for feedback from three different “personas”—which, as a writer, made me think: “So I don’t HAVE to bug three different friends? I can just INVENT them?” The answer is yes, AND you can even invent “friends” who are expert in things your actual friends know nothing about. #VeryCool
Obviously, we are still experimenting with AI and learning how to harness its potential, even as its potential grows exponentially. Like many AI experts, Mollick believes AI will take over some of our jobs and lead us to invent other ones. Who knows? I’m excited to see what my new “friend” thinks about the outline for my latest script. Ta-ta for now!
PS, Ezra Klein of the New York Times interviewed Ethan Mollick for an hour. Here’s the link.
Friends working in NJ are no doubt aware that NJ State Board of Education adopted new NJ Student Learning Standards for ELA on October 4, 2023, and they will take effect this fall.
The revisions, described in NJDOE’s Overview PPT and Crosswalk, include more attention to the Science of Reading and a streamlined, integrated approach to reading, writing, and research. And I must say, I like these revisions. They are sensible, thoughtful, and clear.
I’ve added a new page to the TLC Website called “Standards—NJ,” which includes the NJDOE resources as free downloads and also, for TLC subscribers, an Excel tracker I created.***
Once you and your colleagues take a moment to peruse the updated standards and attendant materials, I encourage you to revisit my approach to UNPACKING standards, explained in this TLC Blog post, which includes a 17-minute video module. The module models unpacking Common Core Standards, but the process is the same.
Last but not least, if you’d like consulting support related to these new standards, please contact me ASAP as I am currently scheduling work for this summer and fall.
***If you’re not already a subscriber, The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
GRAM & GRAN SAVE THE SUMMER by Stephen Chiger and Daniel Pereira
An early blurb for this amazing children’s book states: “Put together the wordplay of The Phantom Tollbooth, the imagination of Willy Wonka and the Chocolate Factory, and the problem-solving mysteries of Encyclopedia Brown, make it about media and digital literacy, and you’ve got Gram & Gran Save the Summer” (Faith Rogow, author of Media Literacy for Young Children: Teaching Beyond the Screen Time Debates).
I could not agree more. And I would add: If an improv comedian and a media literacy teacher had a baby, it would be this book. Educator Stephen Chiger (co-author of Love & Literacy, and full disclosure: a dear friend) and Daniel Pereira (also a former English teacher), along with their illustrator Louis Decrevel, have done something magical. They’ve invented a new genre!
This clever novel takes me back to delicious summer days when I would ride my bike to the library every morning, having devoured the latest Encyclopedia Brown novel the day before, eager to check out another one. This book is playful, hilarious, engaging, and educational all at once. The policy geek in me loves how this book teaches important lessons about how to navigate digital literacy. The English teacher in me appreciates how the authors provide ample context clues for children to build their vocabulary knowledge and strengthen comprehension. And the child in me giggles at the subtle—and sometimes not-so-subtle—humor (yes, one chapter includes extended riffs on flatulence).
Each chapter could stand on its own—and indeed, each one invites rich textual analysis as well as timely discussions about how to spot misinformation on the Internet, how to recognize “cherry-picking” arguments, and more. To support instruction, Steve and Daniel have created a super-helpful Teachers’ Guide.
While they are learning, students will delight in the ongoing story of three children (Nia, D’Angelo, and Deja) visiting their eccentric, unpredictable grandparents for the summer. It’s relatable and a page-turner.
Before I read this book, I didn’t know how much we needed it. And now I want to read a whole series just like it. I can’t wait to see what Steve and Daniel create next!
On the cover, a sallow-faced girl sits half-buried in a ball pit full of yellow balls with smiles drawn on them. She is not playing in the ball pit, however, just staring glumly at her cell phone.
Haidt argues that since 1995, we have 1) overprotected children in the real world and 2) underprotected them in the virtual world. Indeed, this girl could not be more protected indoors–buried amidst a ball pit: a place where your parents have to take you and watch you, a place where you mainly interact with balls, not other children. The balls are soft; it’s impossible to get hurt, even if you fall. And yet this girl is NOT playing, but instead LOOKING AT SOMETHING ONLINE THAT WE (and her parents) CANNOT SEE. It could be anything: text messages from a random stranger, Instagram posts that make her feel inferior, TikTok videos promoting misinformation, pornography….
Overprotective parenting started in the 1990s. Haidt offers several theories for the cause. Perhaps it was an increasing focus on the competitiveness of college admissions, perhaps the rise of cable TV and 24/7 news stories about children being abused by adults, which in turn led to the assumption that adults couldn’t be trusted to be alone with children—so, only their parents could. Whatever the reasons, the consequences have been debilitating. Anxiety levels have risen. Fewer children have close friends. More children feel lonely.
Because unsupervised play in public spaces has evaporated, children have little practice being in “discover mode” as opposed to “defend mode” (p. 67). In “discover mode,” children outside playing freely with other children can learn to take healthy risks, deal with failure, and navigate interpersonal dynamics. They look for opportunities, think for themselves, and see the world like a kid in a candy shop.
By contrast, children in “defend mode” constantly scan for danger and worry about being safe. They operate from a scarcity mindset (p. 71). Ironically, if we could get more children to play outside with one another, we wouldn’t have to do what so many schools now do: try to teach them SEL (social-emotional learning) skills with PowerPoint slides!
Of the many stark graphs that Haidt shares, one of the most dreadful points out the increase in major depression among teens (p. 24). Depression has become 2.5 times more prevalent in girls and boys since 2010. What happened in 2010? For one thing, iPhones came out with front-facing cameras enabling selfies. And we saw a dramatic increase in the number of teens with smartphones. By 2016, 77% of teens owned a smartphone, as did 28% of children between the ages of 8 and 12 (p. 34). In 2012, Facebook bought Instagram, and by 2013, 90 million people were using Instagram to post selfies (p. 35). Because the goal of social media companies is to capture eyeballs (using any means possible) so they can earn advertising income, the rise of social media has led to widespread phone addiction in children (and, let’s face it, adults). For more on this phenomenon, see my post on the documentary “The Social Dilemma.” For so many children, monitoring their phones (esp. their social media) has become like a full-time job. If you’re a teacher or a parent, you know what I’m talking about: these children might be right in front of you, but they feel like they’re somewhere else.
Although densely stuffed with depressing data, this book also offers lots of practical solutions. For immediate action, Haidt proposes these four reforms:
No smartphones before high school.
No social media before 16.
Phone-free schools.
Far more unsupervised play and childhood independence.
Obviously, it’s going to take collaboration to get everyone rowing in the same direction. Regarding #4, if parents can do their part, I would also encourage educators to explore outdoor education opportunities for students. For example, Princeton-Blairstown Center (where, full disclosure, I’ve been a trustee for more than 10 years) offers day trips, overnight trips, and a 5-day Summer Bridge program to help young people (esp. from historically marginalized communities) by nurturing their social-emotional skills through experiential, environmental, and adventure-based programming. Developing these skills enables participants to engage in self-discovery and transform their communities to create a more just world.
Bottom line: Please read this book. Discuss it with your friends and colleagues. Set screen/phone limits with your children. Work with your fellow teachers and administrators to make the difficult but much needed choice of being a phone-free school. Find as many opportunities as you can to get your children, students, and families outside engaged with each other in play. We can work together to turn this thing around.
When an English teacher recently mentioned she was about to launch a research project on famous African-Americans, I asked her for more information. It would be six slides, she said, and they could choose from among 30 people. On each slide, students would answer a question and include an image, then present their answers to the class.
In the past, I would have asked her the following Project Planning Questions to ensure that the work was as meaningful and rigorous as possible:
PROJECT PLANNING QUESTIONS
—What are your objectives? Which skills are students bringing to this project, and which need to be taught? —What questions must students address? Do they begin with HOW/WHY, to ensure that students must EXPLAIN their answers? (When you ask WHAT questions or give students TOPICS, they tend to merely copy and paste information from the Internet.) —How will you teach students to do the research? They must learn to evaluate sources and cite properly. —What are your specific expectations for all aspects of the project? Do you have an exemplar? What is your timeline for each step of the work? How will you ensure students are completing the work efficiently and effectively? How will you provide feedback and support throughout the development process? —How will you evaluate their slides and oral presentation? How will you and their peers provide feedback? How will you hold students accountable for taking notes and/or providing feedback on their peers’ presentations?
Eventually we would get to those questions, but first we needed to consider the elephant in the room: How long would it take AI to complete this project?
We tweaked her questions, then I punched this prompt into Bard (as of 2/8/24, now called Gemini):
“You are an expert on Barack Obama. Create 6 slides on him. Answer these questions: 1) What is this person best known for? 2) How did this person prepare for his accomplishments? 3) What events in his life inspired him to do big things? 4) What are some of his biggest accomplishments? 5) How do you think future generations will remember him, and why? 6) If you could interview him, what are 3 questions you would ask?”
It took about 15 seconds for Bard to generate THIS RESPONSE. The results were fine, but superficial. For example, here’s Slide 1:
Slide 1: The First African American President –Image: Iconic photo of Obama raising his hand after being sworn in. –Text: Barack Obama’s most recognizable achievement: becoming the first African American President of the United States, breaking down racial barriers and inspiring millions.
How could we avoid students using AI to do this very thing? We could not. They have ChatGPT and Bard on their phones and home computers. They could covertly run this prompt and finish their projects in maybe 15 minutes (allowing time for them to find and insert images—and PS, Bard now generates IMAGES, too!), then spend several days in class futzing around, pretending to do work while watching YouTube videos.
Instead, we decided to LEAN INTO the use of AI. Let students use this prompt, I suggested, then push them to dig deeper into the slides by asking HOW/WHY questions about the AI-generated content.
I added this question to that Bard thread:
“For slides 1-5 above, can you please add 2-3 How/Why questions to spark further research about the content on the slide? For example, regarding Slide 4: ‘Why was healthcare reform needed?’ Please rerun slides 1-6 with the new questions added to slides 1-5.”
Again, in 15 seconds, I received THIS USEFUL RESPONSE. (Note: It neglected to include Slide 6, so I had to re-prompt it for that. Here is THAT RESPONSE.)
The teacher now had some decisions to make. The work to answer the questions on any ONE of the slides could be substantial. For example, here’s the revised version of Slide 1:
Slide 1: The First African American President –Image: Iconic photo of Obama raising his hand after being sworn in. –Text: Barack Obama’s most recognizable achievement: becoming the first African American President of the United States, breaking down racial barriers and inspiring millions. —Speaker Notes Barack Obama’s election in 2008 was a watershed moment in American history. It shattered racial barriers and demonstrated the power of hope and inclusion. But his presidency was also about much more than just his identity. Let’s explore the factors that contributed to his historic victory and the significance of his election. —How/Why Questions: What were the key social, political, and economic factors that contributed to Obama’s election?How did Obama’s campaign message of “hope and change” resonate with voters?What is the broader significance of Obama’s election beyond the breaking of racial barriers?
With the newly added questions, students now had a lot to explain, and they would need to sharpen their research skills. AI couldn’t do everything. But it had provided a useful starting point. With that, we turned back to the Project Planning Questions. This was just the beginning!
PS: For more AI support, check out The Literacy Cookbook’s A.I. TIPSpage. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
While some folks argue over whether we should spend more time on phonics or more on comprehension, I have found a way to use AI tools to do BOTH. Because both are needed.
This approach assumes you have phonics instructional materials (e.g., Fundations, Just Words, Orton Gillingham, etc.) and that teachers have been trained in the use of those materials.
To support direct phonics instruction, you also need decodables—stories that utilize the phonics patterns you’ve been teaching (such as short “a” sounds)—so students can practice reading with patterns they now recognize. But often these stories are in short supply.
The GOOD news is that you can use ChatGPT or Bard to create customized decodables.
Here’s a sample prompt to plug into your favorite chatbot:
“You are an expert in phonics. Please write a 150-250 word [realistic/science fiction, you pick] short story using ________ [you pick the phonics sound] for_______- grade students. Please be sure to include at least two characters, a conflict, and dialogue.”
NOTE: You can customize these prompts based on student interests or on content you want to address, making the stories about basketball, weather, volcanos—really, anything!
EXAMPLE: You are an expert in phonics. Please write a 150-250 word realistic fiction short story using long “o” sounds for 4th-grade students. Please be sure to include at least two characters, a conflict, and dialogue.
The EVEN BETTER news is that you can now use Diffit to generate comprehension questions for that same text. Diffit identifies vocabulary words and creates multiple-choice, short answer, and open-ended questions, which you can present to students in the template of your choosing.
Here’s the recipe:
To create your own customized decodable texts, use ChatGPT/Bard with the prompt above. (Or, pull pre-made decodables from the UFLI Toolbox!)
Copy your decodable text and paste it into Diffit here. This will provide vocab and comprehension questions (MC, short answer, open-ended).
In Diffit, select a template you like (there are several dozen). Generate the handout for the students and explain how to work through it. Remember to print an answer key for yourself separately.
NOTE: If you use Google Docs, you can save your decodable texts and Diffit products in your Google Drive, making them easy to share, modify, and re-use.
Try it out! And if you find this post useful, please share it with friends, including educators AND parents.
For more AI support, check out the TLC A.I. TIPSpage. For more resources related to phonics, see the TLC Phonicspage. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
A while back, a friend shared a daily newsletter called “TLDR AI” (“Too Long Didn’t Read AI”). At first I was ecstatic, thinking, I’m gonna be a pro at this stuff in no time. Then I read the first entry. Or should I say: I skimmed the headlines of the first one, because I didn’t even understand the headlines, which promised things like:
RAGatouille library for retrieval pipelines (GitHub repo)
Transitioning from Torch to Jax (12 minute read)
Image Generation with Sequence-Aware Loss (17 minute read)
PowerInfer for LLMs on Consumer GPUS (12 minute read)
I know what a library is. I know what pipelines are. I know what “retrieval” means. But what is a “library for retrieval pipelines”? It went like that, headline after headline, day after day.
Eventually I fell into this routine: I will spend one minute every day skimming the headlines to see if I understand what ANY of them mean. Ninety-nine percent of the time, I don’t. But when I do, I attempt to read that article. And everyone once in a while, I find something useful.
It includes six strategies and various accompanying tactics to improve the ways you prompt AI chatbots (whether you use ChatGPT, Bard, or Bing, etc.) to get whatever you need.
You’re welcome.
PS: For more AI support, check out The Literacy Cookbook’s A.I. TIPSpage. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
If you haven’t seen it yet, check out Parlay Genie, an A.I. tool that generates discussion prompts and enables the running of discussions for free on the platform. The directions are simple:
Write a custom topic or share a URL.
Select # of questions and grade level.
Generate a custom discussion prompt.
Run your discussion for free on Parlay.
You can run discussions in person or online. Check out this “Sample Written Roundtable” demo!
I want to offer a HUGE shoutout to A.J. Juliani, who shared a post about this tool on his super-helpful group “Learning with AI Tools”!!!
PS: For more AI support, check out The Literacy Cookbook’s A.I. TIPS page. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
[This post originally appeared in slightly different form on MiddleWeb on 8/8/23.]
As we look ahead to a new school year with A.I. tools widely available, many teachers are wondering not just “How can I prevent my students from using A.I. to cheat?” but “How can I use A.I. tools to enhance my instruction?”
I have some good news: A.I. chatbots such as ChatGPT and Bard can DRAMATICALLY strengthen writing instruction. What? Yes. Let me explain.
Let’s first acknowledge that it is difficult to teach someone how to write. Ideally, you would share exemplars and non-exemplars of the writing task and have students analyze the differences between them in order to infer what makes the exemplars effective. But who has time to create all of those models? Instead, teachers often ASSIGN writing (“Write a five-paragraph essay on ___!”), offer brief organizational guidance on what to include (“An intro with a thesis!” and “Three body paragraphs!” and “Don’t forget your conclusion!”), then sit back and brace themselves.
The result is predictable: many students struggle, and the teacher doesn’t have time to meet with them all individually. Students become frustrated and announce that they “hate writing.” The teacher, equally frustrated and stressed, isn’t sure what to do.
Some teachers go the other way and OVER-scaffold. They give students fill-in-the-blank introduction paragraphs (“to help get you started”), then write the first body paragraph “together,” and then are surprised when students are reluctant to even begin the next body paragraph on their own.
This sounds terrible. And I thought you were going to explain how to use A.I. to help me?
Yes. Let’s go back to that ideal scenario, in which we provide exemplars and non-exemplars. Chatbots like ChatGPT and Bard can create as many pairs of text as we desire, customized to your heart’s content—by grade level, genre, topic, writing skill (or lack thereof), grammatical issue, and so on.
Cool! What would A.I.-supported writing instruction look like, then?
Here’s a suggested lesson sequence:
MINI-LESSON: Analyze the genre by reading a few A.I.-generated exemplars. Discuss the purposes of this genre (e.g., inform, persuade, inspire, entertain). Ask students: “What do we know about this genre? What do we think are some key elements of this genre?”
MINI-LESSON: Compare A.I.-generated exemplars vs. non-exemplars. Ask: “Now what do we think are key elements of this genre? What makes the exemplars good?”
Here’s an example* for argumentative paragraph writing:
EXEMPLAR
NON-EXEMPLAR
School uniforms should be mandatory for all students in order to promote a positive learning environment. Research studies have shown that schools with uniform policies tend to have lower rates of bullying and peer pressure. For example, a study conducted by Smith and Johnson (2018) found that schools with uniform policies reported a 30% decrease in bullying incidents compared to schools without such policies. When students wear the same uniforms, it becomes harder to judge or discriminate against others based on their clothing choices. Uniforms create a sense of unity among students, as everyone looks equal and belongs to the same community. This reduces the likelihood of students being targeted or excluded because of their clothing preferences, resulting in a safer and more inclusive learning environment for all. By implementing mandatory school uniforms, we can help ensure that every student feels accepted and respected, fostering an environment where they can focus on their education and personal growth without unnecessary distractions or fear of judgment. This enables them to develop important skills, build relationships, and thrive academically.
School uniforms should be mandatory for all students because they help create a positive learning environment. Studies show that schools with uniform policies have lower rates of bullying and peer pressure. By implementing mandatory school uniforms, we can promote a safer and more inclusive environment for students. This will allow them to focus on their education and personal growth without distractions. It’s important for students to feel accepted and respected in order to thrive academically.
*Source: ChatGPT 3.5 (Prompt: “You’re an expert in writing effective paragraphs. Write an exemplar and a non-exemplar argumentative paragraph at the 6th grade level. Include a claim/argument, evidence, explanation, and punchy insight in the exemplar. Leave explanation out of the non-exemplar.”)
Asking “What do you notice?” as students read these texts would lead to a discussion about key components of effective argumentative paragraphs. The exemplar has an argument/claim, evidence, explanation, and a punchy, insightful conclusion. The non-exemplar lacks explanation. Students can quickly see and discuss the value of EXPLAINING in their writing.
3. Students write their own drafts, attempting to include the key elements they noticed in the exemplars.
4. Students pair up and engage in the PARTNER FEEDBACK PROTOCOL. The writer reads aloud while sitting side-by-side with their partner, who does only these three things: 1) Praise specific things they like about the writing, 2) Ask clarifying questions, and 3) Ask the writer to elaborate. The writer does not answer the questions directly but instead makes notes to answer these questions later (in their revision), then keeps on reading aloud.
5. COLLECT WORK AND REVIEW TRENDS TO PREPARE THE NEXT MINI-LESSON. Use A.I. to generate customized exemplars and non-exemplars to support needed revision work. For example, if students are not including relevant evidence, provide paragraphs that include relevant evidence to be compared with those offer random/irrelevant “evidence.”
6. MINI-LESSON: Compare these customized exemplars vs. non-exemplars. Students look at their own writing to see how it mirrors the exemplars or non-exemplars. Ask: “What do you see in the exemplars that you want to do more effectively in your own writing?”
7. Students revise their drafts.
8. While some students engage in the PARTNER FEEDBACK PROTOCOL, conduct WRITING CONFERENCES with small groups of students. (For more info on the Partner Feedback Protocol and Writing Conferences, see this TLC Blog post.)
9. BASED ON OBSERVATIONS OF STUDENT WORK SO FAR, select grammar points to teach. Use A.I. to generate T charts like the one below with exemplary/non-exemplary sentences to help students zero in on key points.
10. MINI-LESSON: Compare those exemplars vs. non-exemplars to bring the grammar points home. In the case below, you could prompt students with this: “The sentences in Column 1 include adjectives, which are underlined; the sentences in Column 2 do not. What does an adjective do? Why do you think good writers use adjectives?” Then ask: “What do you notice in YOUR writing that needs to be addressed?”
COLUMN 1
COLUMN 2
1. The fluffy kitten played with a shiny red ball.
1. The kitten played with a ball.
2. The tall oak tree provided shade on a hot summer day.
2. The tree provided shade on a summer day.
3. My new, colorful backpack has many pockets.
3. My backpack has pockets.
4. The delicious chocolate cake melted in my mouth.
4. The cake melted in my mouth.
5. The fast, sleek sports car zoomed down the highway.
5. The sports car zoomed down the highway.
*Source: ChatGPT 3.5 (Prompt: “You are an expert on grammar and writing. Create a 2-column chart in which 5 sentences in Column 1 use adjectives that are underlined and 5 sentences in Column 2 use the same sentences without the adjectives. Write these at 4th grade level.”)
11. Students revise/edit and submit final drafts.
12. PROVIDE FEEDBACK FOR FUTURE WRITING. Assuming students have not fully mastered the genre, what should they keep in mind for next time? Note: You can use A.I. to create the rubric and to provide feedback with suggested next steps; just be sure to evaluate the A.I. outputs to ensure they accomplish your aims.
13. Create another assignment in which students write a new piece in the same genre. A.I. can help you generate topics/prompts for this new assignment.
14. REPEAT THE CYCLE.
The learning possibilities are endless! And the time saved? Priceless.
PS: For more A.I. support, check out The Literacy Cookbook’s A.I. TIPS page. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
In case you haven’t seen the updated A.I. TIPSpage on The Literacy Cookbook Website, here’s a preview of the ChatGPT page. As you’ll see, it answers key questions and includes brief demos.
What is ChatGPT?
According to ChatGPT itself (Version 3.5, July 20, 2023):
ChatGPT is an AI-powered conversational agent developed by OpenAI. It is part of the GPT (Generative Pre-trained Transformer) series, with GPT-3.5 being one of the versions. GPT-3.5 is based on the GPT-3 architecture, which stands for Generative Pre-trained Transformer 3.
GPT-3.5 is a language model trained on a massive dataset from various sources on the internet, allowing it to understand and generate human-like text. It can perform a wide range of natural language processing tasks, including answering questions, generating text, providing explanations, language translation, and more.
The “Chat” in ChatGPT signifies its primary function as a conversational AI, designed to interact with users in a human-like manner. Users can input text-based queries or prompts, and ChatGPT responds with coherent and contextually appropriate answers.
The model’s ability to handle diverse conversational scenarios has made it a popular tool for applications like chatbots, virtual assistants, language translation, content generation, and various other natural language processing tasks. However, it’s essential to note that while ChatGPT can produce impressive responses, it also has limitations, and its outputs should be carefully evaluated for accuracy and reliability.
NOTE: I’d add that because ChatGPT emerged first, it has gotten the most attention, but Bard (Google’s LLM) now (7-25-23) has more useful features! So take every suggestion on this page and apply it in Bard.
How do we sign up?
Here’s the link: https://chat.openai.com/ NOTE: According to OpenAI’s terms of use, “You must be at least 13 years old to use the Services. If you are under 18 you must have your parent or legal guardian’s permission to use the Services.”
ChatGPT 3.5 is free. There’s a charge for version 4.0.
How can ChatGPT save us time? When prompted effectively, in a matter of seconds, ChatGPT can generate responses to requests for explanations, descriptions, summaries, rubrics, lesson plans, unit plans, translations, and texts of various genres (and reading levels), to name just a few things. It can also use a rubric to provide feedback on writing. This review of The A.I. Classroom provides a list of additional outputs.
Watch THIS VIDEO for a 4-min demo of ChatGPT saving time (creating decodable texts to support phonics instruction).
How can we use ChatGPT effectively in the classroom?
I could write a book on this topic! In the meantime, here are a few suggestions:
PROMPT ENGINEERING: Teach students how to craft effective prompts and evaluate the results. Be careful! As noted below, ChatGPT sometimes provides inaccurate information.
Check out “Mastering Prompt Engineering or the Art of AI Whispering” by Sahbi Chaieb, 3/5/23. This piece discusses the importance of prompt engineering, provides practical examples of prompt engineering, and offers resources and tools for prompt engineering.
MODEL BRAINSTORMING/OUTLINING: Preparation is a key step in the writing process. ChatGPT can generate an infinite number of examples.
DESIGN MINI-LESSONS BASED ON EXEMPLARS AND NON-EXEMPLARS: When asked to compare and contrast exemplars and non-exemplars, students must use analytic skills. Then you can direct them to apply those lessons learned in their own work. For example, if they notice that exemplar paragraphs include claim/argument, evidence, explanation, and punchy insight WHILE the non-exemplars lack explanation OR merely repeat the claim for their conclusion, THEN they can revise their own paragraphs to include the needed elements. Watch THIS VIDEO for a 5-min demo of this approach.
USE RUBRICS TO EVALUATE WORK:
Students can use rubrics and numerous exemplars/non-exemplars to practice evaluating work.
Teachers can use rubrics to score student work.
ChatGPT can use rubrics to generate narrative feedback on student work.
In THIS VIDEO, I demo how to create a writing rubric and use it to generate feedback on a student’s narrative.
DESIGN CREATIVE PROJECTS: ChatGPT can design projects for students to complete. WATCH “How to Save Hours as a Teacher Using A.I. Tools“ by A.J. Juliani, Learning with A.I. Tools, 4/24/23. Stunning 10-min demo of how to create a 6th-grade history/science project and rubric and materials in 2 minutes!
OTHER RESOURCES ON HOW TO USE CHAT/AI EFFECTIVELY:
How can we share ChatGPT responses? In the left-hand bar where all of your prompts are listed, next to each prompt, you’ll see an icon of open box with an arrow pointing out the top; click on that icon and follow the directions.
NOTE: For guidance on HOW TO CITE generative AI tools, go here.
Any cautionary notes?
1. The data used to train ChatGPT came from “various sources on the Internet” up till Sept. 2021. So it is not current. And it can be biased.
2. ChatGPT sometimes provides inaccurate information and sounds very confident while doing so. Don’t be fooled by these “hallucinations.” Check out: “Can ChatGPT-4 Help with History Research? I Think I’ll Pass” by John Norton, MiddleWeb, 3/18/23. This is why it’s so important to evaluate and double-check all results carefully!
3. The creators of ChatGPT (OpenAI) are being sued for various reasons:
PS: For more AI support, check out The Literacy Cookbook’s A.I. TIPS page. The Literacy Cookbook Website offers 2,000-plus teacher-friendly tools. As a bonus for TLC Blog followers, here is the 50%-off discount code for yearlong access to The Literacy Cookbook Website: TLCBOOK50 (Note: ALL CAPS).
The Literacy Cookbook blog 