Predicting AI “misconduct”

A recent headline in the Higher Education section of The Guardian said “Thousands of UK university students caught cheating using AI.” I could see that coming!

It is as if some headlines (taglines, tweets and chyrons) are ready and waiting for a report, evidence base, study or authorised opinion to make them real. To help test the ease of “predictive headlining” I have tried to generate a menu of such headlines ready and on call for the moment.

As I am on holiday, I decided to “cheat” in this task by drawing on an LLM to generate unsettling headlines about both student and staff uses of AI. I’m currently staying somewhere without WiFi. So I decided to use PrivateLLM, an offline AI language system described in a previous post: Local AI.

The local model

I am using a model called “StableLM 2 Zephyr 1.6B.” The model has a context window size of 4,096 tokens (approximately 3,000 words). The model, inference engine and interface reside entirely on my smartphone and require no communication with a remote server.

So, I can provide prompts with abandon and receive any number of responses without incurring network charges, or feeling that I am imposing processing load on some multi-user centralised system. PrivateLLM responses are very fast, but sometimes inaccurate, less functional and less wide ranging than ChatGPT 4.o or LLMs able to access web search, etc.

A menu of headlines

Here are twelve PrivateLLM-generated headlines focussing on student “misconduct.”

• Unrealistic Grades: Concerns Over AI-Powered University Education
• Teaching to the Future? Students and Teachers Warn of Unpredictable, Automated Grading in High-Tech Universities
• AI-Powered Higher Education: Student Body Seeks Transparency as Grade-generating Software Evolves 
• Future Shock: University Students Express Worry About Unregulated Use of Artificial Intelligence in Academia
• Concerns Grow: Allowing AI to Determine Grades Raises Ethical Dilemma for Universities
• Genetic Bias: Some Students See Unequal Treatment When AI Selectively Pours [?] Ingrades [?]
• Students Warn: Expectations Will Be Disrupted by AI-powered Assessments
• Fears Rise: Universities Turning Towards AI to Grade Tests, But What Happens when Machines are Given Human Feelings? 
• Innovative Methods? Unbridled Power of AI to Determine Grades Raises Ethical Questions
• Unseen Power: Universities Questioning the Value of Human Interaction with AI-assisted grading systems
• Some Professors Warn: Automated Grading May Not Reduce Costs or Errors; Instead it Could Lead to Moral Panic and Learning Losses
• Anxiety Builds: Students Confront Future Uncertainty Amid Concerns over Automated Assessment Systems

I also asked PrivateLLM to generate a dozen headlines focussed more on teacher issues.

• AI and Plagiarism: Universities Warn of ‘Language Barriers’ in Preventing Student’s Academic Misuse
• Translating the Future? AI Translates but at Risk of Bleeding [??] Authenticity to Falsehoods
• Student Beware: AI-generated content may pose risks of academic misconduct, plagiarism or hacking 
• Plagiarizing is a New World Order: AI-assisted Translation can’t guarantee the Accuracy and Originality of Student work 
• Lament for the Language Barrier: AI translations of documents may not be as faithful as we expect them to be
• AI-powered Translation Risks Misleading Students with Shoddy, Low-quality Output, Unwittingly Substituting Originality with Copycatting
• AI Translation: Universities Raise concerns about originality, authenticity, and the integrity of student work 
• When Words Fail: AI translation creates language barriers that leave students struggling with the ever-changing world 
• The AI Language Barrier: Universities Warn against using AI to translate students’ work without human oversight 
• Unleashing the Machine: Universities warn against AI-generated plagiarized essays, creating language barriers for students

PrivateLLM’s focus on translation issues is interesting. Here are a few headlines of my own.

• Students Protest Over Poor Grading and Feedback of Their Work That Lacks Teacher Input and Oversight
• Learners Ask Why Are We Paying Universities for What We Can Learn at Home From AI Tutors
• Prospective Students Reject Training for Jobs That No Longer Exist

Counter-headlining

Headlines are generally of a sensational nature to catch attention. Here are a few positive but less catchy headlines, human generated.

• University Encourages Responsible Uses of AI Language Models to Enhance Student Learning
• Standards Rise as Students and Teachers Endeavour to Outcompete Language Models
• Students and Teachers Co-develop New Ways of Working with AI
• Learners and Instructors Develop Well-Informed Critiques of Emerging AI Technologies and Practices.

Reference

• Goodier, Michael. “Revealed: Thousands of UK university students caught cheating using AI.” The Guardian, 15 June, 2025. Accessed 19 June 2025. https://www.theguardian.com/education/2025/jun/15/thousands-of-uk-university-students-caught-cheating-using-ai-artificial-intelligence-survey

Note

• Featured image is from ChatGPT: “Generate newspaper printing machines, with folding and stacking apparatus and with conveyor belts in a large well lit newspaper printing works. Machinery looks disused.”