World

Despite Cheating Fears, Schools Repeal ChatGPT Bans

For decades, Walla Walla High School in the wheat basket of Washington State has maintained an old red wooden barn on campus where students learn a venerable farming skill: how to raise pigs and sheep.

Now, as the new academic year starts, some teachers at the school are preparing to help students learn the latest digital skill: how to navigate A.I. chatbots like ChatGPT.

This month, Walla Walla Public Schools, which serves some 5,500 students, held a daylong workshop on the A.I. chatbots, which can generate homework essays, fictional stories and other texts. About 100 local educators showed up at the high school for the event.

Dozens of local educators recently gathered at Walla Walla High School for a workshop on how to use A.I. chatbots for lesson planning and student learning.Credit…Ricardo Nagaoka for The New York Times

It was a remarkable turnaround for a district that had blocked student access to ChatGPT on school devices only in February.

“I do want students to learn to use it,” said Yazmin Bahena, a dual-language middle school social studies teacher. “They are going to grow up in a world where this is the norm.”

The media furor over chatbots last winter upended school districts and universities across the United States. The tools, which are trained on vast databases of digital texts, use artificial intelligence to manufacture written responses to user prompts. The bots also liberally make stuff up.

Tech giants and billionaires promised that the A.I. tools would revolutionize learning. Critics warned the bots would be more likely to undermine education, inundating students with misinformation and facilitating widespread cheating.

Amid the forecasts of imminent marvels and doom, some public schools tried to hit the pause button to give administrators time to catch up. In December, the Los Angeles Unified School District, the nation’s second largest school system, blocked ChatGPT on school Wi-Fi and district-owned student devices. Other districts soon followed, including New York City, the largest U.S. school system.

But administrators quickly realized the bot bans were ineffective. For one thing, wealthier students who owned smartphones or laptops could simply access ChatGPT, a chatbot developed by OpenAI of San Francisco, or similar bots like Google’s Bard, at home.

“Children who have devices and unfiltered, unfettered connectivity at home are already benefiting from access to these tools,” Alberto M. Carvalho, the superintendent of the Los Angeles Unified School District, said in an interview this week. “Students who depend on district devices and connectivity are restricted.”

In May, New York City schools issued a public mea culpa, saying the district had acted too hastily and would unblock ChatGPT. This week, Mr. Carvalho said that Los Angeles schools were also working on a more permissive policy.

As schools reopen for fall, educators and district leaders are wrestling with complex questions posed by the A.I. tools: What should writing assignments look like in an era when students can simply employ chatbots to generate prose for them? How can schools, teachers and students use the bots effectively and creatively? Does it still count as cheating if a student asks a bot to fabricate a rough draft that they then rewrite themselves?

Some large districts, including Milwaukee, still have ChatGPT blocks in place. Some districts like Newark Public Schools are trying out specialized chatbots specifically designed for student tutoring.

Other districts are embracing tools like ChatGPT as lesson-planning aids for teachers — and as opportunities for students to learn how bots can concoct misinformation and replicate human biases. Administrators say they are simply taking a pragmatic view: Students will need to learn how to prompt chatbots to answer their questions, just as they learn to query search engines like Google.

Teachers at the A.I. workshop studied potential gender and political biases in chatbot texts.Credit…Ricardo Nagaoka for The New York Times

“The world our kids are inheriting is going to be full of A.I. and we need to make sure they are well equipped for it, both the benefits and the drawbacks,” Wade Smith, the superintendent of Walla Walla Public Schools, said in a recent interview. “Putting our heads behind the curtain or under the sheets and hoping it goes away is simply not reality.”

Walla Walla offers a portrait of one district’s remarkable learning curve on A.I. this year. School administrators sought to take advantage of the chatbots’ potential benefits while working to tackle thorny issues like cheating, misinformation and potential risks to student privacy.

In January, Keith Ross, the school district’s director of technology and information services, began hearing about ChatGPT. District teachers were starting to notice a few students submitting chatbot-produced homework as their own. One obvious tip-off: The chatbots fabricated quotes that were not in the novels assigned in class.

The district was also concerned about student privacy. ChatGPT and Bard require new users to provide personal data such as their email address and mobile number. But administrators did not know how the A.I. companies might use students’ account details or their text interactions with the chatbots.

“We just didn’t know enough about the technology,” said Mr. Ross, who blocked students’ access to ChatGPT in February. “We blocked it to buy us some time to get up to speed on what it is and how we were going to support teachers, and potentially students, using it.”

The district set up an A.I. advisory committee with 15 administrators and teachers. The committee studied the potential advantages and challenges of enabling student access to A.I. chatbots and plans to provide more training on the tools for teachers.

“There’s two main categories: using it to be more efficient and save time as a teacher,” said Carrie LaRoy, the district’s technology integration specialist, who helps oversee the committee, “but then also how to teach our students to use it responsibly and with fidelity.”

At 8 a.m. on a recent Thursday, about 100 local teachers and principals trooped into a glass-walled meeting hall at Wa-Hi, as the high school is known. They were giving up a late-summer vacation day to try out A.I. tools for lesson planning and student learning.

The workshop was led by Molly Brinkley, a regional technology trainer who works with 23 local school districts. Most of them blocked ChatGPT last spring, she said.

Some workshop attendees described themselves as chatbot novices. Others said that they had come to pick up more advanced skills.

One of them was Beth Clearman, a veteran honors English teacher at a local middle school who wanted to devise some literary games for the first day of class. So she asked ChatGPT to produce six-word “memoirs” of well-known literary characters.

Molly Brinkley, a regional technology trainer, said schools should reconsider their A.I. chatbot bans, “if teachers receive training.”Credit…Ricardo Nagaoka for The New York Times
Once leery of the A.I. chatbots, Beth Clearman, an honors English teacher at a middle school, is now using them to create writing exercises.Credit…Ricardo Nagaoka for The New York Times

The A.I. chatbot promptly manufactured descriptions like: “lavish parties, unrequited love, green light” and “arrow’s aim, rebellion’s face, Mockingjay’s fire.” Ms. Clearman said she planned to ask students to match the names of protagonists with their chatbot bios. (Spoiler alert: Jay Gatsby, Katniss Everdeen).

Originally leery of A.I. chatbots, Ms. Clearman said she now planned to use ChatGPT “so much!” with her writing students.

“I’ve flipped my whole way of thinking,” she said.

Ms. Bahena, the dual-language social studies teacher, found another potentially useful feature: lesson translation.

“I wanted to see how well it worked in Spanish,” Ms. Bahena said. So she asked ChatGPT to create a quiz on the Civil War in English and Spanish for her eighth grade students. “It did pretty well.”

But even enthusiastic Walla Walla teachers said they were concerned students might have difficulty being sufficiently critical of the materials manufactured by chatbots.

“I’m worried that they might come to take it at face value,” said Shauna Millett, an English teacher at the high school.

Shauna Millett, left, an English teacher at the high school, at the chatbot session.Credit…Ricardo Nagaoka for The New York Times

For now, the district is encouraging teachers to embrace the chatbots, including schooling students on their apparent flaws. Students 13 or older may also create ChatGPT accounts if they wish.

As the workshop wound down, Ms. Brinkley, the regional technology trainer, glanced around the room, pleased to see that dozens of local educators were now comfortable conversing — if not fluent — with A.I. chatbots.

“I do recommend that schools reconsider their bans,” she said, “if teachers receive training, families receive training and students receive training.”

Related Articles

Back to top button