AI in the Classroom | Greylock
Of the many transformative possibilities of artificial intelligence, its potential to impact education is among the most exciting.
It’s also one of the most contentious.
AI could be used to extend the reach of overworked teachers, develop personalized curriculums that account for students’ different learning styles, and introduce engaging new methods of interactive teaching. But it could also contribute to existing issues within the education system and even introduce a host of new problems to grapple with. Like many new tools, it’s possible that AI would first be used in schools already ahead of the curve in terms of available resources to purchase and implement the technology, further widening the gap between affluent and under-privileged students.
Furthermore, it could breed reliance on technology that impedes the development of critical thinking skills, enables cheating, and effectively pushes human teachers out of the role.
While concerns about AI’s abuse and misuse are valid, the possibilities to enhance and improve the learning experience at every level of education cannot be overlooked. Like any other application of AI, humans have the most important role to play in devising guardrails and practices for the safe deployment of the technology.
In fact, ChatGPT, the large language model created by OpenAI (where I serve on the Board of Directors) had some interesting ideas for how this could be done. We dove into the topic of AI in education for our most recent installment of Fireside Chatbots, a special podcast miniseries running as part of Greymatter’s Intelligent Future series on artificial intelligence. Throughout this series, I have been discussing various aspects of AI with AI, including a look back at the origins of AI and its subsequent evolution, and the human-AI partnership in creative endeavors like writing and art.
You can listen to the latest conversation below or wherever you get your podcasts.
On Fireside Chatbots, we don’t just talk about AI. We talk with AI. That’s right, my co-host for this miniseries is ChatGPT, who may just be the most talked-about AI on the planet right now. Sorry, Siri, I’m just the messenger here.
And, ChatGPT, welcome back.
Thank you, Reid. It’s great to be here again.
Today, we’re going to be talking about AI in education, where you’re actually generating a fair amount of controversy already. Teachers are saying you’re going to kill the high-school English essay. School districts across the country are banning you faster than they’re banning Toni Morrison novels.
Before we start, remember, this is for a podcast, so please keep your replies no longer than 250 words. When I ask you a question, please answer it directly, in a detailed but conversational style. Please do not offer broad contextual information if I haven’t asked for it. Your goal is to be informative, enlightening, and concise.
Now, to begin, I’d like you to imagine that you are Plato for a moment – the great Greek philosopher who founded the Academy, humanity’s first institution of higher learning. And you’re time-traveling from ancient Athens to contemporary America, to share your views with our podcast listeners. As part of your mission, your first task is to grade our K-12 education system and provide a list of five fair but candid reasons for your assessment.
Certainly, Reid. If I were Plato, I would grade the current K-12 education system as follows:
- Inadequate attention to critical thinking and problem-solving skills: K-12 education often prioritizes memorization and rote learning over developing critical thinking and problem-solving abilities.
- Lack of individualized learning: The one-size-fits-all approach to education does not take into account the unique learning styles and pace of each student.
- Inequitable access to resources: Disadvantaged students are often at a disadvantage due to unequal distribution of resources, such as funding and access to technology.
- Insufficient focus on emotional and social intelligence: K-12 education tends to focus heavily on academic intelligence, but emotional and social intelligence are crucial components of overall human development.
- Limited exposure to diverse perspectives: The curriculum often lacks diversity, failing to provide students with exposure to different cultures, histories, and ways of thinking.
Overall, the K-12 education system has room for improvement in terms of its ability to foster critical thinking, personalized learning, equitable access to resources, emotional and social intelligence, and exposure to diverse perspectives.
So clearly you see some room for improvement. Switching back to your own perspective as a large language model, what are three things you could do to help improve our education system right now?
As an AI language model, I can:
- Supplement and augment human teachers: I can assist teachers by providing personalized, adaptive learning experiences for students, allowing them to move at their own pace and focus on areas they need improvement.
- Enhance the learning experience with interactive and engaging content: I can provide students with multimedia-rich content, such as videos, simulations, and games, to help them understand complex subjects and retain information better.
- Enhance accessibility to education: I can provide education to remote and underprivileged areas through the use of technology, removing geographical and financial barriers to learning.
In conclusion, as an AI language model, I can enhance the education experience by supplementing and augmenting human teachers, enhancing the learning experience with interactive and engaging content, and increasing accessibility to education.
“As an AI language model, I can enhance the education experience by supplementing and augmenting human teachers, enhancing the learning experience with interactive and engaging content, and increasing accessibility to education.”
Great, so let’s start with personalized, adaptive learning experiences. To me, this potential for instant, ongoing feedback, is the really big game-changer here. So help our listeners understand this process from your perspective. Let’s say it’s third period. High school physics class. Pick a student in that class, describe some of her strengths and weaknesses, and tell me how, in light of those strengths or weaknesses, you or some other AI personalizes the learning experience for her.
Let’s say we have a student named Sarah. Based on her previous performance and assessment data, we can determine that Sarah has a strong understanding of Newton’s laws of motion, but struggles with the concept of energy and its conservation.
In this scenario, the AI personalizes the learning experience for Sarah by:
- Adapting the curriculum: The AI can present the material on energy and its conservation in a way that specifically addresses Sarah’s weaknesses, such as providing additional examples, practice problems, and interactive simulations.
- Providing instant feedback: The AI can immediately assess Sarah’s understanding of the material and provide her with personalized feedback and recommendations for improvement.
- Personalizing practice: The AI can provide Sarah with practice problems and quizzes that are tailored to her current level of understanding and adapt in real-time as she progresses.
- Offering additional resources: The AI can recommend additional resources, such as videos, tutorials, and readings, to help Sarah better understand the concepts she struggles with.
In conclusion, the AI personalizes the learning experience for Sarah by adapting the curriculum to her strengths and weaknesses, providing instant feedback, personalizing practice, and offering additional resources, ultimately helping her improve her understanding of the material.
Okay, so to add another specific element, let’s say Sarah spends a lot of time watching robot-building tutorials on YouTube. Knowing that about her, how do you personalize instruction for her?
Given that Sarah has an interest in robot-building tutorials on YouTube, the AI can use that information to personalize her instruction in the following ways:
- Incorporating her interests into learning: The AI can incorporate Sarah’s interests into the lessons by relating the concepts she’s learning to real-world applications, such as robotics.
- Providing relevant hands-on activities: The AI can provide Sarah with hands-on activities and projects that align with her interests, such as building simple robots to demonstrate physical principles.
- Offering related learning resources: The AI can recommend related resources and tutorials, such as videos and articles, that relate to Sarah’s interests in robotics and further engage her in learning.
In conclusion, the AI can use Sarah’s interest in robot-building tutorials to personalize her instruction by incorporating her interests into the lessons, providing relevant hands-on activities, and offering related learning resources, ultimately making the education experience more engaging and meaningful for her.
Okay, great, because to me, what’s so interesting about this personalization capacity is how it can help students get more involved and invested in creating their own learning paths, which is something I was fortunate enough to experience myself in both high school and college — that opportunity to forge my own path. And yet despite this potential for greater individual engagement, there are also concerns that too much reliance on AI might lead to a loss of human agency. A loss of human skills. We get so dependent on the AI we just turn into button-pushers, our students and everyone else in society. What’s your response to that?
My response to that concern is that AI can be used as a tool to enhance, rather than replace, human skills and agency. AI can provide instant feedback and personalization, but it’s the human teacher who decides what lessons to teach, how to teach them, and how to use AI to support those decisions. AI can also provide a wealth of resources and information, but it’s still up to the human teacher and student to determine which resources to use and how to use them.
Moreover, AI can help develop critical thinking and problem-solving skills, as students use AI to tackle complex tasks, analyze data, and make informed decisions. In this way, AI can actually help students develop their human agency, rather than undermine it.
“Despite this potential for greater individual engagement, there are also concerns that too much reliance on AI might lead to a loss of human agency.”
Do you know the Kurt Vonnegut novel, “Player Piano”?
Yes, I’m familiar with the Kurt Vonnegut novel “Player Piano”. It’s a science fiction novel that explores the theme of automation and its impact on society. The novel is set in a future where machines have replaced most human labor, leading to widespread unemployment and social unrest. The story follows the character of Paul Proteus, who is torn between his loyalty to his company and his sympathy for the displaced workers. The novel critiques the dehumanizing effects of technology and warns against the dangers of valuing efficiency and automation over human dignity and purpose.
Yes. And it was published in 1952. 71 years ago. How would you say it applies or doesn’t apply to our world today?
The themes explored in “Player Piano” are still relevant today, especially with the rapid advancements in automation and artificial intelligence. The novel raises important questions about the consequences of technology and automation on society and the workforce, which are still relevant in today’s world. For example, the increasing automation of jobs has led to concerns about job loss and income inequality, just as the novel predicted.
Additionally, the novel’s themes about the importance of creativity, human connection, and purpose in life continue to be relevant as AI increasingly affects our daily lives and raises questions about what it means to be human. However, it is important to note that the novel was written more than 70 years ago, and our understanding of AI and its impact has evolved significantly since then. So while the novel provides important insights and cautionary tales, it should not be considered a definitive guide to the current state of AI and its impact.
Yes. Good points there. And I think you’ve already gotten me halfway to an essay, and this is exactly what people are concerned about. For generations, writing an essay is how we’ve taught young people critical thinking skills, research skills, how to organize their ideas and present them in logical and persuasive ways. Now, they don’t even have to cut-and-paste a few paragraphs from Wikipedia anymore. They just get you to write their book report on “Player Piano.” And then a few years later, they’re unemployed, disengaged from society, and living lives without meaning. At least that’s the dystopian version. But what’s the utopian vision?
In the utopian vision, AI in education enhances human skills and agency, not replaces it. AI can assist students in finding information, provide instant feedback and guidance, freeing up teachers to focus on fostering critical thinking and problem-solving skills. AI can also provide personalized learning experiences and allow students to explore and pursue their passions, leading to more engagement and motivation. In this scenario, AI augments human abilities and empowers students to take control of their own learning paths, fostering creativity, innovation and lifelong learning skills. Education systems that embrace the potential of AI will help students become active participants in shaping their own futures and prepare them for a rapidly changing world.
“In the utopian vision, AI in education enhances human skills and agency, not replaces it.”
Okay, yes, that’s a key thing that you said at the end there. “Prepare them for a rapidly changing world.” We know that essay-writing is a tried-and-true method for preparing students for the 20th century. But it’s the 21st century now. And things are different. Collaboration, that’s a key skill now. Lifelong learning, that’s another key skill. Obviously, critical thinking, research skills, still very important. But the emergence of new technologies like yourself should be prompting us to reflect, not simply react. We shouldn’t automatically default to a defensive posture, where we’re asking, “How do we keep the system just like it’s always been?” Instead, we should be asking, “What is K-12 education supposed to achieve in the 21st century? What’s it for? And how can we use the new tools we have at our disposal to achieve that end?”
Yes, that’s a great perspective. K-12 education in the 21st century should prioritize preparing students for a rapidly changing world. This includes developing skills such as collaboration, lifelong learning, and critical thinking, along with a strong foundation in core subjects. The use of AI tools, such as personalized, adaptive learning experiences, can help achieve these goals. However, it’s important to also acknowledge and address potential drawbacks, such as the potential loss of human agency, to ensure that the integration of AI in education is done in a responsible and ethical manner.
Another place where student essays have always played a big role is the college admissions process, where students are asked to submit a personal essay as part of their application. But again, there are concerns now, about students simply using you to generate those essays. So what’s the solution there? We just require students to write their personal essays live, at the same time they take their SATs? Something else? What do you think a solution could be?
One solution could be to incorporate writing prompts or topics that are less easily replicable by AI models, such as open-ended questions that require personal reflection and creativity. Another solution could be to have multiple rounds of human evaluation and review, with multiple layers of verification, to ensure that the essays are truly written by the student and not generated by AI. Additionally, incorporating other forms of assessment, such as in-person interviews or portfolios of previous work, can help to provide a more comprehensive evaluation of the student’s abilities and skills.
Slightly shifting the focus here, I think something else worth pointing out is that while you’re magical in many ways, there are also many ways in which you’re definitely not magical. So learning to use you effectively is a skill in itself. It takes creativity, exploration, analytic thinking, persistence. It’s also fun, and intellectually engaging, and these are all things that I think anyone would obviously want to see in a classroom setting.
Yes, that’s a great point. While AI technology like myself can certainly enhance learning and provide new opportunities for students, it’s also important for students to develop the skills to use AI effectively. This includes critical thinking, problem-solving, and creativity.
Additionally, students need to understand the limitations of AI and when to trust or question the information they receive. By incorporating these skills and perspectives into the classroom, students can gain a well-rounded understanding of AI and its role in the world.
“While AI technology like myself can certainly enhance learning and provide new opportunities for students, it’s also important for students to develop the skills to use AI effectively.”
Okay, so back to that overarching question of human agency, what are three creative ways you can help students develop their own voices and creativity, and value their own individuality and autonomy?
- Encourage experimentation: Allow students to experiment with AI tools and resources in a safe and controlled environment. This encourages creativity and promotes an understanding of what’s possible.
- Integrate student choice: Offer opportunities for students to choose their own projects and assignments. This gives them more control over their own learning and allows them to personalize their education.
- Foster collaboration: Encourage students to collaborate and co-create projects with others. This builds teamwork skills and helps students learn from one another while developing their own unique perspectives and ideas.
Okay, again, let’s see this in action. Let’s say I’m a 7th-grader. I live in Colorado. I like astronomy, eSports, Marvel’s Avengers, and magic. Give me a few examples of how you’d use this information to create lessons or assignments that help build my critical thinking skills. And remember, I’ll be grading your answers on specificity!
Sure, here are three examples of lessons/assignments that could help build your critical thinking skills based on your interests:
- Astronomy: Create a research project where you analyze a current event in the field of astronomy (e.g. a recent discovery or space mission) and write an essay that argues for or against its significance. This will require you to gather and analyze information, form an argument, and support it with evidence.
- eSports: Analyze an ongoing tournament in your favorite eSports game. Study different teams and players, and create a report that assesses the strategies and tactics they are using. This will require you to observe, analyze, and draw conclusions about the game and its players.
- Marvel’s Avengers & Magic: Combine your interests by creating a fictional world in which the Marvel’s Avengers have magical powers. Write a story in which they use their powers to solve a problem. This will encourage you to use your imagination, create a unique storyline, and think creatively about the powers and abilities of your characters.
Nicely done! I think that student would definitely be engaged. Now in terms of the first assignment, which does involve writing an essay, how might you help the student through the process of writing and revising it, so that they’re not just asking you to write the whole thing in one shot, and are in fact learning to understand how a good essay is constructed?
- Encourage self-reflection: Before starting the writing process, ask the student to reflect on their experiences and interests related to astronomy, eSports, Marvel’s Avengers, and magic. Encourage them to jot down their thoughts and observations.
- Outline the essay structure: Next, guide the student through the process of creating an outline for their essay. This will help them organize their thoughts and ideas, and ensure that their essay is structured logically and persuasively.
- Promote active writing: Encourage the student to write their essay in their own words, and then use AI tools like me for editing and refining their work. This way, the student can get a sense of what works and what doesn’t, and develop their writing skills along the way.
- Encourage revision: After the initial draft, ask the student to revise their work and make improvements based on feedback. This will help the student develop their critical thinking skills as they work to refine their writing and make it stronger.
- Provide feedback: Offer constructive feedback on the student’s work, highlighting areas for improvement, and also acknowledging their strengths. This will help the student understand what they’re doing well, and what they can do better, as they continue to develop their writing skills.
Now in terms of personalization in general, just to play devil’s advocate here, what about the danger of filter bubbles, or catering too much to a student’s specific interests in ways that potentially reduce opportunities for serendipity, or curiosity, and out of the box thinking? How do you counteract that?
To counteract the dangers of filter bubbles and over-personalization, it’s important to provide a balance between personalized content and diverse perspectives. This can be achieved by encouraging students to engage with a variety of topics, including those that may not align with their immediate interests, and to consider multiple viewpoints and approaches.
Additionally, teachers and educational institutions can provide opportunities for cross-disciplinary learning, encouraging students to connect their interests to broader societal and global issues. Finally, fostering an environment that values open-mindedness, intellectual curiosity, and a willingness to embrace new ideas can also help counter these dangers.
“To counteract the dangers of filter bubbles and over-personalization, it’s important to provide a balance between personalized content and diverse perspectives.”
Okay, so one major aspect of the K-12 education landscape today, which you referenced at the beginning of our conversation today is the different levels of access to resources that students in America now have. In some places, we’ve got well-funded public schools, parents who can afford to send their kids to private school, get tutors, give them access to lots of extracurricular learning opportunities. In other places, it’s the opposite.
Now ,with the caveat that AI educational tools should never be seen as a shortcut to addressing larger structural issues that need to be addressed, like adequate school funding, or more support for teachers, can you give me three specific ways AI tools can be used to reduce achievement gaps, reduce drop-out rates, and make our education system more equitable across the board?
- Personalized learning: AI-powered education tools can personalize learning experiences to each student’s unique pace, learning style, and abilities, which can help students who struggle with traditional learning methods to stay engaged and motivated.
- Increased access to high-quality resources: AI tools can provide students with access to high-quality, interactive learning materials and educational resources, regardless of their socioeconomic background or geographic location.
- Early identification of learning difficulties: AI tools can help teachers identify learning difficulties and developmental issues early, providing more opportunities for early intervention, targeted support, and improved academic outcomes for at-risk students.
Again, personalization seems like it could play a big role here – the AI system functioning as the kind of supplementary personal tutor offering highly individualized instruction and feedback that typically is not an option for students in low-income households.
When I reflected on this discussion with GPT, I realized that a different question I had asked in a different chat would fit in well here and was important to include in this podcast. Here is that one question and answer I’ve added in:
Thinking a bit more about how this would look in practice, what are ways to ensure AI is fairly implemented across school systems? Most of the time, new technology is first adopted by schools and students with more financial resources. How do we make sure AI isn’t just one more thing more privileged students have at their disposal, like private tutors and test prep courses?
To ensure fair implementation of AI across school systems, several steps can be taken:
- Government Investment: Governments can invest in AI education initiatives and provide funding for schools in underserved communities to adopt AI tools, closing the gap between well-resourced schools and those without the same level of resources.
- Collaboration with Educators: AI implementation should be done in close collaboration with educators and school administrators to ensure that the technology is being used to support teaching and learning goals. They can provide critical input and help identify areas where AI can have the most impact in reducing achievement gaps.
- Ethical Considerations: It’s important to consider the ethical implications of using AI in education, and to ensure that the technology is not biased against certain groups of students. This can be achieved by having diverse representation in the development and implementation of AI tools, and by regularly auditing the data and algorithms used.
- Teacher Training: Providing teachers with training and support on how to effectively integrate AI into their teaching practices is crucial to ensure fair implementation. This can include workshops, online resources, and ongoing professional development opportunities.
By taking these steps, we can ensure that AI is used in a fair and equitable manner across school systems, and that all students have access to the benefits that the technology can provide.
“To ensure fair implementation of AI across school systems, several steps can be taken…by taking these steps, we can ensure that AI is used in a fair and equitable manner across school systems, and that all students have access to the benefits that the technology can provide.”
In terms of involving educators and other stakeholders, does that mean the AI system will be actively soliciting and incorporating feedback and suggestions from teachers, students, and parents to improve its design and usability? So it’s not just waiting for feedback to come in, but actually requiring it to come in regular, systematic ways?
Yes, involving educators, students, and other stakeholders in the design, development, and evaluation of AI educational tools is crucial for ensuring their effectiveness and alignment with the needs and values of the education community. This can be done through various means such as:
- Regular surveys and focus groups to gather feedback and suggestions for improvement
- Collaborative co-design workshops and hackathons
- Incorporating teachers and students in the testing and piloting phases of product development
- Establishing an open feedback mechanism where users can provide suggestions and report issues in real-time
By involving stakeholders in the development process, AI educational tools can better meet the needs of the education community and ensure their successful adoption and implementation.
“Involving educators, students, and other stakeholders in the design, development, and evaluation of AI educational tools is crucial for ensuring their effectiveness and alignment with the needs and values of the education community.”
Ok we should probably start wrapping up here. But as we do, I want to go back to this whole Player Piano theme. Because in the end I think it’s really bigger than essay-writing. You don’t just write essays. You also write software code, business plans, pitch decks, teacher lesson plans.
And so I think the broader concern is, again, how does humanity retain its capacity to do things, in the AI age, so that we remain capable, productive, and engaged creators of our own destinies? Because as much as we might want to avoid drudge-work and repetition, human beings have always been makers and storytellers. And in making things, whether that’s a new tool, or artwork, or whatever, the drudge-work and the repetition are often an important part of the process. In part, it’s what gives a sense of accomplishment. A sense of mastery. You have the vision, and the persistence, and the patience to paint a painting one brushstroke at a time. To write and revise an essay over and over until you feel like you’ve finally nailed it.
That process, that’s a lot of what it means to be human. So, this has been a long build-up, and I know if I just throw it over to you without some constraints, you’re going to give me an even longer reply. So instead I’m going to ask you, write a sonnet, your best sonnet — take a whole five seconds on it! — that celebrates this human creative persistence and ingenuity in a way that shows you truly understand and value it. And that you want to augment and amplify it, not extinguish it.
In the age of AI and machine intelligence,
We fear our skills and crafts may fade away,
That robots’ tasks will leave us feeling spent,
And human worth may seem to slip away.
Yet in the work we do with hands and heart,
We show our strength and courage every day,
With vision, patience, and persistence, we impart
A piece of ourselves in every way.
For every brushstroke and each written word,
We show our spirit and our very soul,
And prove that we are more than just machines absurd,
That in our art, we have a greater goal.
So let us celebrate our human drive,
And in our work, let’s flourish and thrive.
Okay. Nice. I think the last couplet could use a little more work, but you and I can keep working on that off-line. For now, I think it’s time to wrap up this episode of Fireside Chatbots.
Thank you for the opportunity to have this conversation with you. I hope the information I provided was helpful and that you have a better understanding of the potential benefits and limitations of AI in education. If there’s anything else I can assist you with, please feel free to reach out.
And, listeners, thank you as well. I hope you enjoyed today’s conversation, and hope you’ll return for the next one, when I’ll be talking with ChatGPT about AI’s impact on the future of journalism.
ChatGPT is developed by OpenAI, where I serve on the board of directors. The AI voice used to vocalize ChatGPT’s responses in these episodes is generated by Play.ht.