Why AI is forcing universities to rethink the way they run exams
A last-minute decision to return to pen and paper has put a spotlight on higher education's struggle to ensure academic integrity in the age of AI, writes Catherine McGregor in today's extract from The Bulletin.
Handwriting makes a surprise comeback for some students
As mentioned in Tuesday's Bulletin, Victoria University of Wellington's law school has announced that it is returning to handwritten exams for some courses this trimester, citing fears of AI-assisted cheating. The sudden decision affects students in two 300-level courses, both of which are externally regulated and require in-person, invigilated assessment.
In a message to students, law dean Geoff McLay said the university lacked a reliable technical solution to prevent unauthorised AI use on laptops. The affected students seem unimpressed, with one telling the Herald it's 'backwards behaviour' that has 'left students in the lurch'. Vic's provost Bryony James has acknowledged the bad timing, but says handwritten exams remain 'the gold standard of ensuring integrity'.
NZ's mixed assessment landscape
While Victoria pushes back against AI in exams, other universities are taking different approaches. At Auckland University, all law exams remain digital, backed by lockdown browsers (which deny access to unauthorised websites) and in-person invigilators. At Waikato it's the opposite: all law exams are handwritten, with exceptions for students with accessibility needs. It's a mixed picture when it comes to other disciplines, too. An Auckland University spokesperson tells RNZ that 61% of all exams were digital in 2024. At Victoria University, it's only 30%.
Among universities considering a return to handwritten (or 'blue book') exams, changes to the format may sometimes be required. According to Bryony James, 'one way to help students faced with handwritten tests [is] to make them multi-choice, or require answers in bullet-point form', RNZ writes.
A warning from the US
Victoria University's decision coincides with an alarming investigation by New York magazine (soft paywall) into rampant AI cheating at US universities. Author James D. Walsh paints a picture of widespread, almost casual reliance on tools like ChatGPT to complete assignments. Sarah, a Canadian university student, explains her dependence on AI bluntly: 'I spend so much time on TikTok. Hours and hours, until my eyes start hurting, which makes it hard to plan and do my schoolwork. With ChatGPT, I can write an essay in two hours that normally takes 12.'
The problem isn't confined to essays – students are now using AI to pass coding assignments, solve complex maths problems and cheat in internship interviews, Walsh writes. Many professors are on the verge of giving up. One 'is now convinced that the humanities, and writing in particular, are quickly becoming an anachronistic art elective like basket-weaving'.
Says the professor: 'Every time I talk to a colleague about this, the same thing comes up: retirement. When can I retire? When can I get out of this? This is not what we signed up for.'
The detection dilemma
While universities are going old-school to beat AI cheating in exams, many are relying on tech solutions to catch students who use AI to write essays at home. Massey University has leaned heavily on Turnitin's AI-detection feature, but its rollout has been mired in controversy. Last April the Sunday Star-Times reported that at least 20 Massey students had 'claimed amnesty' in exchange for confessing to using ChatGPT to cheat. Provost Giselle Byrnes said the university uses Turnitin's AI-detection 'as a last resort' but a number of students said they had been accused of cheating after receiving false positives, the Manawatu Standard reported.
A recent Reddit post from another New Zealand student detailed their struggle with faulty detection tools. Under the headline 'AI has ruined my university experience', the user wrote that they 'truly put a lot of effort into my work [but] recently all of my assignments have been coming back as AI-generated … I'm sick of looking like a cheater, and I know none of my tutors believe me when I say I don't use ai.'
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Otago Daily Times
5 hours ago
- Otago Daily Times
Large retailers support facial recognition technology in stores
By Phil Pennington of RNZ The heads of a dozen of the largest retailers and telcos in the country have come out in strong support of using facial recognition technology in their stores. This follows the Privacy Commissioner giving a "cautious tick" to a trial in New World and Pak'nSave supermarkets. "The undersigned major New Zealand retailers strongly support the use of fair and accurate technology to protect our workers and customers," said a statement at industry group Retail NZ's website. Without saying when they might start using it, they stated they would work now to develop "best practice". "We recognise that technology must be used in a fair and accurate way." The letter was signed by the heads of Briscoes and Rebel Sport, Bunnings and Mitre 10, Michael Hill Jewellers, Farmers and The Warehouse, the two Foodstuffs supermarket groups in the two islands as well as rival Woolworths, and telcos One NZ and Spark. Concern lingers over privacy of shoppers Privacy Commissioner Michael Webster had said his report was "not a green light for more general use of FRT" (facial recognition technology). "However, we recognise the importance of the issue for many businesses." The trial let other businesses ask themselves the right questions about whether to use FRT and in what ways to protect privacy, Webster said. There were significant caveats. "While the percentage of misidentifications may be small, rolling FRT out at scale would mean that large numbers of people would be misidentified." Foodstuffs North Island's own research suggested 900 shoppers a year could be misidentified in its stores alone. The commissioner suggested raising the algorithm accuracy from 90 percent to 92.5 percent, among other measures. A Māori Reference Panel set up at the end of 2024 told the commissioner it opposed FRT's use in supermarkets. This was "given the vital role of supermarkets in providing access to food, the current supermarket duopoly which means there are limited alternative options for people who are barred from entry, and the concern that the whole population of Aotearoa will be subjected to surveillance in supermarkets in order to reduce instances of harmful behaviour by a small minority of customers". How does it work? Retail NZ's Carolyn Young said for someone to be on the watchlist, they had to have offended and/or been abusive and/or aggressive in store and trespassed. If someone was trespassed from a retail environment, they currently are not able to return to that store for two years. "What we know in retail is that recidivous offending is very high - between 35-50 percent (depending on the sector) of offending is done by recidivous offending. "So we know that even though someone has been trespassed, they continue to come back into store," Young said. "FRT will enable stores to identify these individuals as they enter store to ensure that the store is safe for staff and customers. "FRT does not enable customers to be monitored. It takes an image of people as they enter the store and if they are not on the watchlist, then they are deleted immediately. "FRT does not provide ongoing monitoring throughout the store, just one photo/image as someone enters." The big-store signatories said they acknowledged the commissioner's oversight, and Foodstuffs for leading the way with its trial. "The use of FRT in the right settings with the right controls will provide positive benefits and outcomes for customers, retailers and workers, while not impeding on the privacy of New Zealanders. "The vast majority of customers will be able to go about their business as usual and will in fact be safer in those stores where FRT is used," the Retail NZ statement said. Webster also stressed it would be "highly desirable" to do training of the FRT algorithms on New Zealanders' faces, by consent, to cut down the risk of bias and inaccuracy. Very limited such testing by the Department of Internal Affairs last year found the imported tech it is currently using was accurate. What happens overseas? Many multiple FRT systems are on offer that return different rates of accuracy in independent tests by the US-based benchmarking agency. In Australia, Bunnings had been in a legal fight with a watchdog that asserted its facial recognition there impinged on people's privacy. Reports of the tech being used at supermarkets in an isolated way in New Zealand date back to at least 2018. Researchers foresaw the tech spread in 2021. "Private sector use of FRT-enabled surveillance is likely to increase, particularly in the retail sector, especially as these services come 'baked-in' to vendor offerings," their landmark report for police said. That contributed to police deciding not to use FRT on live camera feeds, a constraint they say they have stuck with till now. In Britain, the tech's spread, for example in airports and shopping centres, prompted the government's biometrics ethics group in 2021 to recommend oversight by an independent ethics group including of collaborative FRT use between retail and police. Young said Britain was a long way ahead of New Zealand in terms of the implementation of FRT and had used CCTV actively in the community for many years. Here, the Office of the Privacy Commissioner (OPC) would carry out the role of oversight, she said. "It may be in the future that there is a need for another regulatory body to do this work, but while we are in our infancy of implementation and the OPC has been very clear about how it is to be rolled out, we believe that the parameters for implementation are very clear." The Privacy Commissioner's report does not contain a similar recommendation. It mentioned Foodstuffs auditing how it compiled watchlists of people for the camera-software to look out for, but not that this should be independent. The signed Retail NZ statement did not mention independent overview. Australia's privacy regulator signalled in March it would be proactive in regulating biometric information. Biometrics include face, fingerprint and iris - unique identifiers of who a person is. The Office of the Australian Information Commissioner - it had taken on Bunnings, which was appealing - put this in a wider frame: "Our research told us that more than a quarter of Australians feel that facial recognition technology is one of the biggest privacy risks faced today, and only three percent of Australians think it's fair and reasonable for retailers to require their biometric information when accessing their services". "Thinking about what the law permits, but also what the community would expect" was critical.
Otago Daily Times
5 hours ago
- Otago Daily Times
Facial recognition technology supported by big name retailers
By Phil Pennington of RNZ The heads of a dozen of the largest retailers and telcos in the country have come out in strong support of using facial recognition technology in their stores. This follows the Privacy Commissioner giving a "cautious tick" to a trial in New World and Pak'nSave supermarkets. "The undersigned major New Zealand retailers strongly support the use of fair and accurate technology to protect our workers and customers," said a statement at industry group Retail NZ's website. Without saying when they might start using it, they stated they would work now to develop "best practice". "We recognise that technology must be used in a fair and accurate way." The letter was signed by the heads of Briscoes and Rebel Sport, Bunnings and Mitre 10, Michael Hill Jewellers, Farmers and The Warehouse, the two Foodstuffs supermarket groups in the two islands as well as rival Woolworths, and telcos One NZ and Spark. Concern lingers over privacy of shoppers Privacy Commissioner Michael Webster had said his report was "not a green light for more general use of FRT" (facial recognition technology). "However, we recognise the importance of the issue for many businesses." The trial let other businesses ask themselves the right questions about whether to use FRT and in what ways to protect privacy, Webster said. There were significant caveats. "While the percentage of misidentifications may be small, rolling FRT out at scale would mean that large numbers of people would be misidentified." Foodstuffs North Island's own research suggested 900 shoppers a year could be misidentified in its stores alone. The commissioner suggested raising the algorithm accuracy from 90 percent to 92.5 percent, among other measures. A Māori Reference Panel set up at the end of 2024 told the commissioner it opposed FRT's use in supermarkets. This was "given the vital role of supermarkets in providing access to food, the current supermarket duopoly which means there are limited alternative options for people who are barred from entry, and the concern that the whole population of Aotearoa will be subjected to surveillance in supermarkets in order to reduce instances of harmful behaviour by a small minority of customers". How does it work? Retail NZ's Carolyn Young said for someone to be on the watchlist, they had to have offended and/or been abusive and/or aggressive in store and trespassed. If someone was trespassed from a retail environment, they currently are not able to return to that store for two years. "What we know in retail is that recidivous offending is very high - between 35-50 percent (depending on the sector) of offending is done by recidivous offending. "So we know that even though someone has been trespassed, they continue to come back into store," Young said. "FRT will enable stores to identify these individuals as they enter store to ensure that the store is safe for staff and customers. "FRT does not enable customers to be monitored. It takes an image of people as they enter the store and if they are not on the watchlist, then they are deleted immediately. "FRT does not provide ongoing monitoring throughout the store, just one photo/image as someone enters." The big-store signatories said they acknowledged the commissioner's oversight, and Foodstuffs for leading the way with its trial. "The use of FRT in the right settings with the right controls will provide positive benefits and outcomes for customers, retailers and workers, while not impeding on the privacy of New Zealanders. "The vast majority of customers will be able to go about their business as usual and will in fact be safer in those stores where FRT is used," the Retail NZ statement said. Webster also stressed it would be "highly desirable" to do training of the FRT algorithms on New Zealanders' faces, by consent, to cut down the risk of bias and inaccuracy. Very limited such testing by the Department of Internal Affairs last year found the imported tech it is currently using was accurate. What happens overseas? Many multiple FRT systems are on offer that return different rates of accuracy in independent tests by the US-based benchmarking agency. In Australia, Bunnings had been in a legal fight with a watchdog that asserted its facial recognition there impinged on people's privacy. Reports of the tech being used at supermarkets in an isolated way in New Zealand date back to at least 2018. Researchers foresaw the tech spread in 2021. "Private sector use of FRT-enabled surveillance is likely to increase, particularly in the retail sector, especially as these services come 'baked-in' to vendor offerings," their landmark report for police said. That contributed to police deciding not to use FRT on live camera feeds, a constraint they say they have stuck with till now. In Britain, the tech's spread, for example in airports and shopping centres, prompted the government's biometrics ethics group in 2021 to recommend oversight by an independent ethics group including of collaborative FRT use between retail and police. Young said Britain was a long way ahead of New Zealand in terms of the implementation of FRT and had used CCTV actively in the community for many years. Here, the Office of the Privacy Commissioner (OPC) would carry out the role of oversight, she said. "It may be in the future that there is a need for another regulatory body to do this work, but while we are in our infancy of implementation and the OPC has been very clear about how it is to be rolled out, we believe that the parameters for implementation are very clear." The Privacy Commissioner's report does not contain a similar recommendation. It mentioned Foodstuffs auditing how it compiled watchlists of people for the camera-software to look out for, but not that this should be independent. The signed Retail NZ statement did not mention independent overview. Australia's privacy regulator signalled in March it would be proactive in regulating biometric information. Biometrics include face, fingerprint and iris - unique identifiers of who a person is. The Office of the Australian Information Commissioner - it had taken on Bunnings, which was appealing - put this in a wider frame: "Our research told us that more than a quarter of Australians feel that facial recognition technology is one of the biggest privacy risks faced today, and only three percent of Australians think it's fair and reasonable for retailers to require their biometric information when accessing their services". "Thinking about what the law permits, but also what the community would expect" was critical.
RNZ News
a day ago
- RNZ News
Google turns internet queries into conversations
Photo: Unsplash Google is now letting people turn online searches into conversations, with generative artificial intelligence providing spoken summaries of query results. With Audio Overviews, Gemini AI models quickly sum up query results in conversational style, according to Google. "An audio overview can help you get a lay of the land, offering a convenient, hands-free way to absorb information whether you're multitasking or simply prefer an audio experience," Google said in a blog post. "We display helpful web pages right within the audio player on the search results page so you can easily dive in and learn more." Google is beefing up online search with generative artificial intelligence, embracing AI despite fears for its ad-based business model. CEO Sundar Pichai recently unveiled a new AI mode in Google search. The search engine's nascent AI mode goes further than AI Overviews which display answers to queries from the tech giant's generative AI powers above the traditional blue links to websites and ads. Since Google debuted AI Overviews in search slightly more than a year ago, it has grown to more than 1.5 billion users across several countries, according to Pichai. Google's push into generative AI comes amid intensifying competition with OpenAI's ChatGPT, which has itself incorporated search engine features into its popular chatbot. - AFP
