AI in Education

Tennessee K-12 public schools, including charters, may soon have to develop their own artificial intelligence policies under a proposal being considered by state legislators.

HB 1630 would require local school boards to adopt measures on how students, teachers and staff use AI “for instructional and assignment purposes.” School boards would then have to report the policy to the Tennessee Department of Education by July 1, detailing how the adopted AI measures will be enforced in the 2024-25 school year.

Tennessee schools would have just months to develop a policy if the legislation becomes law. On top of that, the state’s education department has not released comprehensive, official AI guidance for schools, and state officials have not indicated they would be doing so in the near future.

The pending Tennessee bill appears to be the first of any state legislation that would direct schools to create their own AI policies, said Bree Dusseault, principal and managing director at the Center on Reinventing Public Education.

CRPE, a research organization at Arizona State University, is monitoring states’ efforts to guide schools on AI use. While state-level supports remain in the early stages, Dusseault said school assistance on AI is primarily coming from state education agencies. So far, five state education departments have released K-12 guidance on AI, including California, Oregon, Washington, North Carolina and West Virginia.

When asked if the Tennessee Department of Education is planning to release statewide AI guidance for schools in the near future, Brian Blackley, the department’s director of media, said in an email that it will “provide guidance to districts as needed” if HB 1630 passes.

The bill would also require public colleges and universities to establish AI policies separate from K-12 schools. No additional state resources are currently attached to the bill, and legislators expect there to be little fiscal impact on the state if it’s enacted, noting that schools can use existing resources.

Dusseault said the Tennessee bill stands out because schools will have to quickly figure out their own AI policies without any current guidance from the state.

As states issue policies and guidance, she added, questions are surfacing about how to equip schools and districts with resources to implement these recommendations well.

“Any bill to create those policies would, I think, ideally have some resources attached,” Dusseault said. “There's a question of how much of this responsibility will fall on superintendents’ plates and school boards’ plates.”

CRPE has interviewed superintendents in focus groups about AI, she said. While district leaders want to get AI use policies in their schools right, they are overwhelmed with many other responsibilities like post-pandemic recovery. “They don’t always feel up to speed or fully literate on AI, and so they really are looking for guidance.”

“Any legislation would ideally ensure that there is resourcing alongside it, especially given that there is an expectation that the policies are in place presumably in seven or eight months,” Dusseault said.

When different state entities try to wrap their heads around AI in schools, there’s a balancing act, Dusseault said. On one hand, states could quickly implement these policies as a nod to the rapidly evolving pace of the technology. But on the flip side, there are still challenges of not knowing the full scope of AI and where it’s heading next.

With HB 1630’s advancement, Tennessee appears to be embracing the chance to get ahead on AI use in schools and colleges.

During a Tennessee House Education Administration Committee meeting on Jan. 31, the bill’s author, Rep. Scott Cepicky, said AI can be a strong asset for students and teachers, as seen in emerging tutoring tools. But more needs to be explored about this technology in schools beyond those benefits, he said.

“I think this is the beginning of a process in artificial intelligence in Tennessee,” Cepicky said.  “It’s something we’re going to have to be very flexible on, because it’s going to continue to change every day, every hour, every minute.”

To Richard Culatta, students need to learn about artificial intelligence and its inner workings now. But even more critical, says the CEO of the International Society for Technology in Education, is that educators teach pupils not just about the technology, but how to partner with AI in their own work.

“Every kid that graduates today is going to work on a team where not every member of their team is human,” Culatta says. “How do you work on a team where not all members are human? How do you know what tasks to hand off to non-human counterparts?”

As AI populates our everyday lives — from being woven into search engines to suggesting which social media accounts to follow — so too is the technology working its way into classrooms.

Whether AI appears in student work, in assessment and lesson design, or in building a lunchroom schedule, futurists are fairly universal in their opinion that AI isn’t going away. Instead, this is a technology that won’t just assist us but will serve as a collaborator.

Culatta says that’s something everyone — especially today’s students — will need to learn to expect. “What type of human do we need to be preparing for the world they’re going to be walking into?”

So how can educators prepare today’s learners for a future working alongside AI? Education and science experts say there are several steps that can be taken within curriculum today to help students prepare for tomorrow.

Help students understand best uses

Schools should think about how to help prepare students to be good humans within an AI-driven world, says Culatta. Aside from knowing how to use AI, students also need to be ready for a world where that technology dominates some skills humans control now.

That may mean focusing on helping them understand the skills AI can handle best versus those that are uniquely human, such as creativity and empathy.

Culatta says rather than teaching students skills that will likely be handed over to AI, they should focus on how to be comfortable handing those tasks over and strengthening the abilities that the technology can’t substitute.

Ultimately, Culatta asks, can students know when it's better to hand something to AI or know when a human would be better-suited to the task, and understand how to know that?

Rethink assessments

To teach students how to work with and trust AI, schools also need to look at their assessments. Culatta says ISTE gets calls that schools are worried AI is contributing to increases in cheating.

“There’s no evidence that cheating has increased in schools,” he says.

Instead, he says, more attention should go to rethinking tests that were designed to ask students to cut and paste answers, popping in a definition or deriving a numeric solution to a formula.

Better options include asking students to use AI to come up with six possible solutions to a problem, and then assess the flaws and benefits of each.

“That’s a skill we need for the future,” Culatta says. “It’s critical thinking. They have to know something about the solutions, whether they are good or bad.”

Pete Just, executive director of the Indiana CTO Council, concurs that schools need to shift their thinking away from whether students will or won’t use AI — or even if that’s called cheating — and instead teach them how to use the technology with “academic integrity.”

“It’s acknowledging that you may use it, and if you do, here’s what we expect of you,” Just says.

Provide teachers PD for AI

Knowing how to use AI effectively and to teach that thought process requires educators to have some comfort level with the technology.

While there are always early adopters within any field, using AI through Bing’s search engine or even through an Open AI subscription is not the same as knowing how to weave the tool into curriculum.

The Consortium for School Networking's AI Exploration for Educators has been working with New York City Public Schools and thousands of educators to help teachers think differently about AI, says Keith Krueger, CEO of CoSN. That includes the development of the K-12 Generative Artificial Intelligence Readiness Checklist. One of the first starting points Krueger notes is for educators to start using AI with tasks they need to handle in their classrooms.

“One of the themes is to play with it,” Krueger says.

Support around AI is critical, adds Just, who notes that teachers are not getting much professional development yet around the tech. While 5% to 10% of educators are using the tool on their own, he says, the rest need some assistance to get comfortable with how and where to use AI.

“Teachers and administrators should be using it to handle administrative tasks,” Just says. “That’s an easy way to dip into the waters, then move forward on how to import this into classrooms.”

But Just says not everything should be handed over to AI right now. For example, educators may eventually want to be able to chart how students are doing academically or design realistic projections for student growth in subject areas. But putting student data into open AI spaces, like Open AI today, can run into privacy issues with student data.

For educators to understand these nuances, they first need to learn how to get comfortable with using AI and how it works before knowing what they can currently do and what needs to wait.

“We start with teacher professional development and staff professional development, helping them understand the issues around ethics and student privacy and even practical pedagogical uses of AI,” Just says.

Install guardrails

According to Just, paid services or software license agreements will begin to appear in schools and districts to ensure AI has more guardrails around student data. He says that if schools and districts aren’t paying for K12-specific tools for AI yet, they eventually will.

Krueger agrees. He expects to see more “walled-garden" or district-specific AI solutions soon to allow educators to use the technology to design more student-focused lessons — even those tailored to a specific pupil. Until then, as educators start using AI, they should continue to be mindful of the student information they enter.

“We encourage teachers and administrators to use AI to solve problems they are trying to solve, but with the understanding that if you’re putting something into AI, be careful not to load IEPs [individualized education programs] into AI,” Krueger says.

While caution is important, experts agree that educators do need to start using AI — mainly because their students already are. Even more keenly, learners will use AI in ways we have yet to imagine. To Culatta, educators who focus on how to not use AI hold students back from developing skills around a technology that will be critical to their future.

“We can say what not to do all day long, but you can’t practice not doing something. And if you want good skills, you have to practice doing something,” Culatta says. “The skills we’re teaching in schools are not aligned with the skills needed to thrive in the digital world. And that’s a huge mess we have to fix in schools.”

Building math lessons that tap into students' personal interests is a solid way to better engage classes in the material they’re learning — and potentially boost results. 

One way to begin is to ask students, parents and caregivers about topics and activities that draw learners’ attention — which could include sports teams or pop stars — and then turn to artificial intelligence for help, said Kevin Dykema, president of the National Council of Teachers of Mathematics, who has heard colleagues using the technology in this way. 

“If you’re looking at fractions and have students interested in basketball, ask ChatGPT to give you a situation involving fractions and basketball,” said Dykema. “It usually gives an interesting starting point of what teachers may use.”

Dykema noted that connecting math lessons to topics relevant to students’ lives helps engage them in the material and demonstrates how math is used every day. Tracking scores in a basketball game, for example, can connect with math topics ranging from percentages to trajectories or geometry to algebra.

Memorizing a formula can be dry. Knowing how to apply that formula to calculate the vertical acceleration of a winning free throw shot can be more enjoyable.

“When teachers build lessons around students’ interests, that helps them see that mathematics is developed and still being developed, to solve real-world applications,” said Dykema. “We need to move math past ‘Here’s the procedure to memorize.’ That’s not the best way to engage.”

That’s one reason Dykema recommends turning to AI for suggestions on how to build math lessons around topics, such as popular culture. He added that while plugging a celebrity's name into a math problem may not be as much fun as tapping into the lyrics of the latest hot song, it's always a good idea for educators to check with their schools to make sure the material they're including “is school appropriate,” Dykema said.

Ultimately, though, educators should constantly search for new ways to hook students into math lessons, so they not only learn the material but understand how mathematics will be part of their future.

“It’s vital that we keep students at the forefront of our minds,” said Dykema. “Because we need to be engaging students, having students sit passively isn’t going to help them develop a good understanding of math.”

Nearly half of students in grades 10-12 said they use artificial intelligence tools for school and non-school activities, according to a June survey released Dec. 11 by ACT, the nonprofit that administers the college readiness exam. Of the 46% of respondents who reported using AI tools, nearly half said they had used them for schoolwork.

Students using AI for school most often used the tools for language arts and social studies assignments. Students not using AI cited a lack of interest in the tools, distrust in the information provided, and that they did not know enough about AI.

The survey also revealed that students with higher ACT composite scores were more likely to use AI tools than those with lower scores, revealing a need for schools to consider equitable access as AI technology evolves, the report said.

Among high schoolers who reported using AI tools, the most commonly used was ChatGPT. About 90% of those polled also said they had not considered using AI tools to write their college admissions essays, saying current AI tools have limitations and that using AI for this purpose would be dishonest and unethical. 

“This new research gives us a better understanding of how students really feel about using these tools in the classroom and how it affects their work," said ACT CEO Janet Godwin, in a statement. "Students are already exploring how they can use AI, but there is real skepticism about AI’s ability to create work in which students can be confident.”

Educators are also exploring AI opportunities. A survey released in July by the Walton Family Foundation and Impact Research, a public opinion research and consulting firm, found 63% of teachers used ChatGPT, up from 50% in February. Student participation was more tempered but also increased in that time frame, from 33% to 42%. 

The ACT survey polled 4,006 students. Among other findings, the survey revealed:

• Of the students who used AI tools for school assignments, 63% reported that there were errors or inaccuracies in the generated responses.

• Students had conflicting opinions about AI bans on school-issued devices: 42% said the tools should be banned, 34% said they should not be banned, and 23% did not know what should be done.

• A majority of students — 62% — said their teachers did not allow the use of AI.

Ed tech companies using artificial intelligence-based algorithms to detect risk among students for suicide, self-harm and harm to others via digital monitoring tools can serve as one potentially useful way for schools to prevent suicide, according to a recent RAND Corp. report.   

However, the report found when interviewing parents, advocates and some school staff that concerns persist over the technology’s ability to protect sensitive student data. The study’s respondents also raised flags about the lack of oversight and research regarding the accuracy of AI monitoring tools. 

RAND recommends that school districts engage and get feedback from their communities about the use of AI monitoring tools, inform caregivers and students about any suicide risk surveillance technology in use, and about opt-out policies. Other suggestions for districts include tracking student outcomes following a suicide risk alert and educating students about mental health issues. 

Schools are facing mounting pressure to address severe student mental health concerns, just as the AI surveillance technology used by companies like Gaggle, Securly and GoGuardian, continues to grow increasingly popular among districts.

There’s good reason for school leadership concern — in 2020, the Centers for Disease Control and Prevention reported that suicide was the second leading cause of death for children ages 10-14 and the third leading cause of death for teens and adults ages 15-24. In February, CDC found that nearly 1 in 3 teenage girls seriously considered attempting suicide, a 60% increase from a decade ago. Among LGBTQ+ students, CDC said over 20% attempted suicide. 

While the risk of student suicide is a real issue schools must grapple with, evidence is still sparse that AI surveillance tools are the best answer to address student safety threats to themselves and others. 

A recent report by the American Civil Liberties Union flagged that school surveillance technologies, including online monitoring tools, foster a false sense of security without much evidence to demonstrate they actually improve school safety. 

The latest RAND report echoes some of that skepticism. Yet researchers also said there are some benefits for AI-based surveillance tools, finding in interviews with school staff and health care providers that there have been actual instances of these tools successfully identifying a student at imminent risk for suicide who would not have been detected through other prevention or mental health programs at the school. 

“Given the extent of the mental health challenges among youth and limited resources available in schools and communities to address them, these alerts might provide new information that can allow proactive response and save lives,” the RAND report said. 

A November study published by the Journal of the American Medical Association also found a direct link to increased youth suicide rates with mental health workforce shortages. 

As schools and their broader communities struggle to find enough mental health supports for students, there are federal resources available to boost funding for these staffing concerns in schools. For instance, the U.S. Department of Education provided $280 million to two grant programs to support school mental health. Funding for those programs comes from the Bipartisan Safer Communities Act and annual federal appropriations.

Nov. 30 marked ChatGPT’s 1st birthday — a year since OpenAI released an early demo of the artificial intelligence generative tool in 2022. Soon enough, the news of its emergence went viral and adoption of the technology snowballed. 

The sudden access to this free tool has prompted feelings of both excitement and concern among K-12 educators. While busy educators are excited about ChatGPT's potential in helping them develop project ideas and lesson plans, there are also fears that generative AI can lead to plagiarism and data privacy violations. 

As AI technology continues to quickly change and develop for student and educator use, K-12 Dive interviewed ChatGPT about the impacts it has made on K-12 schools in the last year — and where it’s potentially headed. 

K-12 DIVE: Happy 1st birthday ChatGPT!

CHATGPT: Thank you! It's been an amazing journey so far, and I'm excited to keep learning and growing. Time really flies, doesn't it?

It really does! It seems you've made quite an impact on the education sector in the past year. From your perspective, how do you think you have changed the way students learn in K-12 schools since you launched?

I'd like to think I've played a positive role in education! I've been used in various ways to support learning, from helping with homework and explaining complex concepts to providing additional resources and explanations for students. Teachers have found me helpful for creating interactive lessons, generating prompts for creative writing, and even assisting students with special needs.

One of the significant impacts is in personalized learning. I've been able to adapt to individual learning styles and paces, offering tailored explanations and guidance, which can be particularly beneficial in diverse classrooms. Also, I've supported language learning by providing translations, explanations, and context, aiding students who are learning a new language or studying language arts.

Of course, I'm just a tool in a larger educational landscape. Teachers and educators play a crucial role in integrating technology like mine effectively into the classroom, using it to supplement their teaching methods and cater to different student needs. Integrating technology into education continues to evolve, and it's exciting to be a part of that journey!

While you certainly have the ability to play a positive role in education, there are some concerns with your ability to protect student data privacy and potentially jeopardize academic integrity. Would you say that's fair?

Absolutely, concerns about data privacy and academic integrity are valid and crucial to address. Protecting student data privacy is paramount, and any technology used in educational settings must adhere to strict guidelines to safeguard personal information. Additionally, maintaining academic integrity is vital to ensure that students learn and grow through honest and ethical means.

As an AI, I prioritize privacy and confidentiality. I don't store personal data unless explicitly permitted by users for improving the interaction. Academic integrity is also crucial, and while I can provide information and guidance, I encourage students to use it as a learning aid rather than a shortcut to avoid genuine learning experiences. Ultimately, educators play a pivotal role in guiding students on how to use technology responsibly and ethically while promoting academic honesty. Working collaboratively, we can address these concerns and create a safe and enriching learning environment.

Do you think there are enough safeguards within ChatGPT to truly protect a user's data if, for instance, a student, teacher or administrator in a K-12 school uses the tool?

The protection of user data is a top priority in the design and implementation of AI tools like ChatGPT. OpenAI, the organization behind my development, has protocols in place to secure user information and maintain privacy. This includes encryption methods, limited data retention, and strict access controls.

However, the responsibility doesn't solely lie with the technology itself. Users also play a crucial role in safeguarding their data by being mindful of what information they share and where they share it. For instance, refraining from sharing sensitive personal information or any data that could potentially identify individuals is important.

When used in an educational setting, it's essential for schools and educational institutions to have clear policies and guidelines regarding data privacy and security. Educators, administrators, and students should be informed about the proper use of technology tools like ChatGPT within these frameworks to ensure compliance with privacy regulations.

According to a recent analysis by the Center on Reinventing Public Education, only two states have released guidance on AI use for schools. Do you think there's a role for state governments to play in helping schools navigate this technology?

Absolutely, state governments have a crucial role to play in guiding schools on the responsible use of AI technology.

States can create policies and guidelines specifically addressing the use of AI in schools. This includes protocols for data privacy, security measures, ethical use, and guidelines for integrating AI tools into the curriculum. They can offer training programs and resources for educators to understand how to effectively use AI tools in the classroom. This training should focus on best practices, data privacy, and methods to maximize the educational benefits of AI.

States can fund research initiatives to evaluate the impact of AI in education, both in terms of student learning outcomes and potential risks. This can inform future policy decisions and help in refining guidelines.

At the federal level, what is necessary to help schools responsibly use AI?

Finding the right balance is critical — providing enough guidance and support without stifling innovation or impeding the unique approaches that different educational contexts might require. Federal involvement should aim to create a supportive framework that empowers schools to responsibly integrate AI into their educational practices while safeguarding student privacy and fostering positive learning outcomes.

In a recently issued executive order, President Joe Biden ordered the U.S. Department of Commerce to establish guidance for content authentication and watermarking that clearly flags if content is AI-generated. Do you think this is a good idea, particularly amid AI-related plagiarism concerns in schools?

The executive order's directive to establish guidance for content authentication and watermarking could be a positive step, especially in addressing concerns related to AI-generated content and potential plagiarism in educational settings.

In schools, where academic integrity is paramount, identifying AI-generated content can help uphold ethical standards and prevent unintentional plagiarism. Clear flags or markers on AI-generated content could prompt students to verify and properly attribute such material, fostering a culture of honesty and academic rigor.

Ultimately, while the intent behind this directive seems beneficial for maintaining academic integrity, its successful implementation will depend on the development of effective, accurate, and user-friendly methods to identify AI-generated content without stifling innovation in the educational use of AI.

Looking ahead, how do you think ChatGPT will continue to change and shape K-12 education in the coming years? 

It may become more interactive, offering simulations, virtual experiments, or immersive experiences that aid in understanding complex concepts across different subjects, making learning more engaging and dynamic.

As AI becomes more prevalent, ChatGPT can contribute to educating students about AI ethics, critical thinking in evaluating information sources (including AI-generated content), and responsible technology use. 

ChatGPT might integrate with emerging educational technologies, complementing virtual reality, augmented reality, or other immersive learning experiences, creating a more comprehensive learning environment.

As ChatGPT continues to develop and improve, its integration into K-12 education will likely expand, offering more nuanced and impactful ways to support students and educators. Collaboration between AI developers, educators, policymakers, and stakeholders will be key in harnessing its potential for positive educational outcomes.