Soaring levels of cheating using AI on exams and coursework at many UK universities. Cases of AI misuse have, on average, trebled over the last year at the universities which responded to our request for data. Well, Newsnight has spoken anonymously to some of the cheating students about how they use artificial intelligence.

I 100% use it in some way on every assignment I’ve got. So, nobody I’ve known firsthand has ever been caught. But, for everything, even assignments, exams, because most of my exams are pretty much all at home, open-book. So, people just copy and paste, and just put that in. But even if we were in a class and they’re like, “Oh, discuss this” and stuff, people would just look up the question on chat GPT. And then read off the screen and pretend they’ve been discussing with their groups.

I mean, genuinely, I think everybody’s lost the ability to think for themselves.

If you loaded up a multiple choice test on one of your modules, you could just, like, get out snipping tool, or whatever. Basically, select the question and then that copies the image on and then you could just be, like, “What’s the answer to this?” And then 5 seconds later, it will give you the right choice. So, it’s very, very seamless, very quick to do. It’s just so easy and it doesn’t really require any thinking whatsoever to get an answer.

I kind of realized in my second year, early second year, that I basically didn’t have to do my readings full stop … … and I would still get a high first throughout second and third year, even though I wouldn’t engage with the seminar material at all.

We probably should’ve got AI to voice that up. So, what’s the solution? Dr. Edward Howell is a lecturer in International Relations at the University of Oxford.

I fundamentally do not believe that telling students not to use AI will do anything. We have no way of tracing, of checking, whether they are using AI or not. What we need to do, though, is to go back to handwritten examinations because that at least generates a level playing field … …and, after all, that’s what a university is about, teaching students how to think.

So, now let’s talk to a man who has made a career out of catching students cheating. Chris Cameron is the CEO of the plagiarism detection company, Turn It In. Good evening to you, Chris. How often, then, do you catch AI cheaters?

Well, so, let me just quickly say the main goal of our company is to deter students from cheating in the first place. And, typically, when schools use our software, the rate of misconduct drops in half. Students just don’t cut the corner in the first place.

To your question, right now we’re seeing about 20% of student essays run through our system with meaningful amount of AI in them. And, roughly, I’d say 10% almost completely written by AI.

And so, just explain to us how it works. You use AI to catch AI?

That’s right, yes. We basically took student essays written pre-AI. We had large language models, like chat GPT, write essays to those same assignments and then we had an AI train itself. We said these 10,000 essays are from a human. These 10,000 are from AI. You figure out how to determine what is organic and what is inorganic.

Now, I suppose a cynic might suggest (and it works remarkably well!) it’s in your interest to talk up this problem. Some universities have told us that AI detectors themselves are unreliable and they can throw up false positives.

Yeah. I mean, AI by it … by its nature makes mistakes. In our case, it’s very low. It’s about one in 200 students would be falsely accused of AI misconduct. That’s the bad news. The good news is getting to the truth is very simple. If a student says, “You told me I used too much AI and I really didn’t,”… … all the student has to do is share their Microsoft Word Doc or Google Doc with a teacher. You can look at the revision history and see very quickly whether the student wrote this over a number of hours with typical typing patterns, … … or whether it was a big giant paste of an essay from Chat GPT.

So, if that protocol is followed, you should never have a student falsely penalized for misusing AI.

And how does the UK compare to other countries in terms of dealing with this problem?

Yeah, it’s a really good question. So, I’d say, historically, the UK is very progressive in using technology like ours. It was the first country in the world, 20 years ago, that had widespread use of our anti-plagiarism solution. This is pre-AI. And very rapid adoption of new technologies, like online marking, etcetera. With AI, it’s an anomaly because most of the rest of the world has, like, 80% adoption of AI detectors.

You mentioned in the UK it’s like 15%. I think that’s a real loss because when you have our technology in place and students know about it, … … they’re incented to not cut the corner, to do the real work, to learn … to learn how to communicate, think critically. So, I really hope the UK has a change of heart across its universities and moves from where it is today, which is very low usage compared to rest of world, to on par.

Chris Cameron, thank you for joining us. Let me tell you what the universities themselves have said. Universities UK told Newsnight: “Academic misconduct, including passing of others work as your own, can result in serious penalties. But we can’t just see this technology as something to be avoided and afraid of. It is here and we have to work out how to support our students to use it well and appropriately.”
(End)

• To soar (verb): to increase rapidly.
• Coursework (noun): work or tasks set at regular periods as part of an educational course.
• Misuse (noun): using something in an incorrect way.
• Anonymously (adverb): in a way where the identity of the person is unknown.
• An assignment (noun): work or a task given to students, such as homework or coursework.
• Firsthand (adverb): directly or from personal experience.
• Open-book (adjective): allowing students to use their notes to complete a task.
• To pretend (verb): to speak or act in a way that gives a false impression.
• Genuinely (adverb): honestly, in a way that tells the truth.
• To load up something (phrasal verb): to load and open a programme or doc on a computer.
• A snipping tool (noun): a tool on a computer that allows you to select certain information.
• Seamless (adjective): smooth and simple, without interruptions.
• Whatsoever (adverb): in any way at all.
• Full stop (idiom): at all, completely.
• A first (-class honours degree) (noun): the highest grade possible at a British university.
• To engage with something (verb): to take part or interact with something.
• A seminar (noun): a class for university students in which different topics are discussed.
• A solution (noun): an answer to a problem.
• Fundamentally (adverb): in the most basic and important way.
• To trace something (verb): to find out where something comes from.
• A level playing field (idiom): a situation where everyone is equal, nobody has an advantage.
• After all (idiom): anyway, in any case.
• Plagiarism (noun): copying another person’s work or ideas and presenting it as your own.
• To deter someone (verb): to try to discourage or prevent someone from doing something.
• In the first place (idiom): in the beginning.
• A rate (noun): the speed of change or progress of something.
• Misconduct (noun): poor behaviour that is considered to break the rules.
• To cut a corner (idiom): to not do something fully, to take an inappropriate shortcut.
• Meaningful (adjective): significant or important.
• Roughly (adverb): approximately, more or less; without precision.
• Organic (adjective): developing in a natural or slow and gradual way.
• A cynic (noun): a person who criticises something or does not completely believe it.
• To talk up something (phrasal verb): to talk about something in a very positive way.
• To throw up something (phrasal verb): to produce or reveal something new or unexpected.
• A false positive (noun): a result of a test that shows something as present, which is untrue.
• By (something’s) nature (phrase): naturally or inherently.
• To accuse someone (verb): to say that someone has broken the rules or committed a crime.
• Typical (adjective): normal, average.
• A protocol (noun): a set of rules or instructions for doing something.
• Progressive (adjective): promoting progress towards new ideas, policies, or methods.
• Widespread (adjective): popular, common, or used across a large area.
• Adoption (noun): accepting and starting to use new ideas and methods.
• An anomaly (noun): something unusual that doesn’t follow the normal rule or pattern.
• To incent(ivize) something (verb, US non-standard): make a person want to do something.
• Critically (adverb): in a way that carefully considers the advantages and disadvantages.
• A change of heart (idiom): a change of opinion, belief, or decision.
• To be on par (idiom): to be at the same level or equal to something or someone else.

In this text, there are examples of the zero, first, and second conditional. Conditionals describe the result of a certain condition or situation. The if/ when clause tells you the condition, the main clause tells you the result. The order of the clauses can be changed, and this doesn’t affect the meaning.

Zero Conditional: used for general truths, laws, or rules – things that are always true.

Structure: if/when + present simple, + present simple:

‘If you touch fire, it burns.’    ‘Water boils if it reaches 100°C.’   ‘When you smile, other people smile.’

First Conditional: used for real or possible future situations – things that are probable or likely to happen.

Structure: if/when + present simple, + will (or other future form) + infinitive:

‘If I’m hungry, I’ll order a pizza.’   ‘She’ll call when she arrives.’   ‘If it rains, we’re going to stay inside.’

Second Conditional: used to imagine unreal, unlikely, or hypothetical (imaginary) situations in the present or future – things that are not likely to happen or impossible.

Structure: if + past simple, + would (‘d) + infinitive:

‘If I had wings, I would fly everywhere.’   ‘He’d be happier if he lived by the sea.’   ‘I’d help if I could.’

With the verb ‘to be’, ‘if I/ she/ she/ it + ‘were’ is grammatically correct; but ‘if I/ she/ she/ it + was’  is common in informal speech.

‘If I were/ was taller, I’d play basketball.’    ‘If he were/ was rich, he’d buy a yacht.’   ‘We’d sleep better if it were/was quieter!’

*Other words and phrases like unless, even if, or in case can replace if/ when in conditional sentences:

‘I’ll go unless you need me here.’   ‘We’d win the match even if the other team had 20 players!’   ‘Here’s my number in case you need it.’

*Modal verbs, such as could, should, might, may can also be used in the result clause (not the if clause):

‘If you study hard, you might pass the exam.’   ‘If I had a lot of money, I could buy a bigger house.’

Examples in the text:

• But even if we were in a class …, people would just look up the question on chat GPT: the second conditional (‘even if + past simple, would + infinitive’) is used here to talk about an imaginary situation in a university class.
• If you loaded up a multiple choice test …, you could … get out snipping tool: the second conditional, using the modal verb ‘could’ in place of ‘would’, is used here to talk about an imaginary possibility using AI.
• … when schools use our software, the rate of misconduct drops in half: the zero conditional (‘when + present simple, + present simple’) is used here to talk about a general truth or result.
• If a student says, “You told me I used too much AI and I really didn’t,” all the student has to do is share their Microsoft Word doc or Google Doc with a teacher: the zero conditional is used here (with a cleft ‘all …’ structure) to talk about a general truth, what students can do to prove their innocence if they are suspected of cheating.
• … if that protocol is followed, you should never have a student…: a passive form of the first conditional (if + present simple passive, + future form), using the modal verb ‘should’ to mean ‘is likely’, is used here to talk about the probable future results of using the AI cheating protocol.
• … when you have our technology in place…, they’re incented to not cut the corner: the zero conditional is used here to talk about a general truth or result when using the technology.

https://learnenglish.britishcouncil.org/grammar/b1-b2-grammar/conditionals-zero-first-second

#AI #Education #Universities #BBCNewsnight

http://englishpodcastscollected.com/new/wp-content/uploads/2026/01/BBC-Newsnight-Students-and-AI-cheating-everybodys-lost-the-ability-to-think-for-themselves.pdf

http://englishpodcastscollected.com/new/wp-content/uploads/2026/01/BBC-Newsnight-Students-and-AI-cheating-everybodys-lost-the-ability-to-think-for-themselves.docx