Teachers at Tauranga Boys’ College can use ChatGPT artificial intelligence to generate school report comments. 

A college document - obtained by the Bay of Plenty Times and titled “2023 Senior End-Of-Year Reports” - addresses subject teachers about the option of using the artificial intelligence chatbot. 

“You are welcome to use the integrated ChatGPT, which will generate a comment based on Achievement, Learning Behaviours and Attendance,” the document says. 

“However, you MUST check these carefully, ensuring the information is correct, has UK English spelling, and no more than 500 characters.” 

The document includes a photo of a textbox with a green button saying, “ChatGPT: Generate Comment” and says “peer-checking comments” is required. 

Using ChatGPT for school reports causes concern 

A source, who spoke to the Bay of Plenty Times on condition of anonymity about the use of ChatGPT for school reports, said in their view: “It would be nice to have proper feedback.” 

Tauranga Boys’ College trialled using ChatGPT for report comments last year and received "mixed feedback". Photo / Alex Cairns 

The source said the school says ‘‘it’s based on your achievements and your behaviour and things like that, but [from my perspective] it would just be nice to have some actual proper feedback, I guess.” 

The source said ChatGPT appeared to collect data “so [in my opinion] this could very well be a privacy concern as well”. 

Tauranga Boys’ College responds 

In a statement, Tauranga Boys’ College principal Andrew Turner said ChatGPT was added to the nationwide KAMAR student management system - which stores student information - last year “for all schools to trial”. 

“This information is strictly confidential to the school and key information can be accessed by staff, students, and parents via login/password access.” 

Turner said the ChatGPT feature synced with the school’s KAMAR information and was “private and confidential”. 

Tauranga Boys' College principal Andrew Turner. Photo / Mead Norton 

He said the feature collated students’ academic achievement, learning behaviours, and attendance data and provided an “initial outline comment” for staff to use based on the information. 

“This feature aims to improve accuracy and efficiency in reporting but never take away the staff’s voice when reporting to students and parents.” 

Turner said the school expected staff to “peer review” their reports within their departments. Reports were then proofread by the administration team and senior leaders. 

He said staff trialled the tool with “mixed feedback”. 

“We are not convinced this is the way forward but remain open to exploring tools that will support staff in improving their accuracy and efficiency when reporting to students and parents.” 

KAMAR responds to ChatGPT concerns 

In a statement, KAMAR chief executive Kent Lendrum said its student management system was used by schools nationwide and served as a school’s “main student database” for administration tasks, such as attendance and reporting. 

Lendrum said KAMAR provided interfaces to integrate with third-party tools, including ChatGPT. 

However, “we do not collect information on what tools each client uses”. 

Lendrum said it was up to schools to decide if and how they used each tool, ensuring it met their policies, procedures, and safeguards. 

Lendrum said it identified some teachers were “already using” ChatGPT to assist them with tasks such as writing report comments. 

“Adding it to KAMAR made this new technology available to all our schools - only if they configured it for their staff to use - and added a layer of privacy by using a generic account, standardising the requests and not sharing identifying information such as the student’s name or gender.” 

Lendrum said teachers should always check comments for “coherence and completeness” to ensure they were an accurate reflection of the student. 

Lendrum said KAMAR did not collect any information on which schools had configured and enabled ChatGPT. 

Ministry of Education says ChatGPT can be ‘unreliable’ 

Ministry of Education Curriculum Centre acting hautū (leader) Pauline Cleaver said schools were not required to inform officials if they were using ChatGPT for report comments. 

The ministry acknowledged ChatGPT had “distinct potential benefits” for schools because it could synthesise large amounts of data. 

“However, they can be unreliable and can invent facts and details.” 

The ministry recommends schools ensure information is verified by teachers who know the student and can recognise errors. 

Cleaver said the ministry’s advice and guidance regarding AI was that schools were expected to have “robust policies and procedures around privacy, digital technology use, and data information security”. 

Privacy Commissioner: Schools should let parents know of plans to use ChatGPT 

A spokeswoman from the Office of the Privacy Commissioner said schools should give parents an opportunity “to talk through potential benefits and concerns” of using ChatGPT for report comments. 

“If the school decides to go ahead, they could let parents who remain uncomfortable opt out of using AI for their children’s reports.” 

Schools should also take steps to mitigate privacy risks by completing a Privacy Impact Assessment, the spokeswoman said. 

“This can help parents and schools feel more comfortable that a new technology is being used safely.” 

The spokeswoman said organisations had legal responsibilities under the Privacy Act to look after personal information. 

“Using a fake name for a student may not be enough to keep it safe. The information is still about that student, and when it’s shared on to third parties it may be possible to re-identify the student or use the information in unexpected ways.” 

“Information about children is sensitive and could have lifelong effects. We would like to see those risks being considered before using this kind of service.” 

The spokeswoman said ChatGPT was “still new” and there was “a lot we don’t know”. 

Privacy Commissioner Michael Webster last May released advice on the expectations of agencies and businesses about AI to respect people’s privacy rights. 

This included having senior leadership approval, reviewing whether AI was “necessary and proportionate,” conducting a privacy impact assessment, being “transparent” with customers and clients about using it, developing procedures about accuracy and access by individuals, ensuring “human review” before acting and ensuring personal or confidential information was not retained or disclosed by the AI tool. 

Megan Wilson is a health and general news reporter for the Bay of Plenty Times and Rotorua Daily Post. She has been a journalist since 2021.