I'm on the other end of this, failing students for using AI to write their papers.
As one of my colleagues put it: "If you couldn't be bothered writing it, why should I be bothered marking it?"
Oh sweet baby Jesus no, burn AI to the ground, humanity can't be trusted with it
- Thread starter TheMysteriousGX
- Start date
Recommended Videos
As one of my colleagues put it: "If you couldn't be bothered writing it, why should I be bothered marking it?"
The funny thing is, they still spend quite a lot of time actually wrangling the AI to come up with something. Maybe still not as much as writing it themselves, but hours for just a few pages. And yet I assume they take away a lot less from just spanking an AI into doing what they want.
That said, how are you checking for AI? My sister who taught high school ran into this problem when she realized that AI detecting software isn't perfect by any means.
AI is pretty much the death of the written essay as an assessment.
Mostly Universities use a software submission platform such as Turnitin. Turnitin's big benefit is that has an automatic plagiarism checker, but it has also added an automatic AI checker last few years. No AI checkers are that good. The biggest problem with AI checkers are going to pick up false positives if students have used some form of assistance software like Grammarly. The language style of AIs is very plain: precise, clean, short sentences, no spelling and grammar errors, just like Grammarly will convert writing into (I mean the basic Grammarly - there's also an advanced form which has generative AI). It may even flag a student who has developed a superior academic writing style. What it can be most useful for is a screening, "first pass" for things to scrutinise more closely.
So then we examine more closely with academic judgement, and there are various criteria we can use to consider. For academic judgement, most universities by now will provide criteria to check by. Some might have these available. You can look around at some suggestions of typical AI habits (e.g. Wikipedia has one) but good luck finding the time to do that level of analysis and getting a good enough case. Even if we were that good at spotting AI - and we're not - the time it takes is likely unsustainable in workload.Most students cheating with AI are smart enough to not make major errors like also copy and pasting the AI prompts and other major giveaways. Fake academic references that don't exist is usually one of the most common giveaways I've encountered that make for a strong case, but that's more a university thing (school projects don't tend to have referencing). Sudden major changes in writing style or quality if you have an example of a student's previous submission is a potential sign.
And then with any accusation of cheating, the standard in most universities is civil law, "balance of probabilities". But that student is going to take a punishment which could have substantial repercussions, such as their degree classification, or even withdrawal. If a student really wants to fight this, you need to think whether the university will uphold it, and then potentially the law. So... how confident are you the court will agree your decision was right? How about the uni, paying all the cost of the case, plus potentially a payout if it loses? Therefore in practice the uni will lean to the student's side. There are plenty where I think there is a strong suspicion the student has used AI, but it's not going to stick so we don't pursue it.
Therefore In practice the majority of students who cheat with AI are at least moderately careful will get away with it. (They might get poor marks, of course.)
My online learning thing I'm currently made to do uses AI, and describes itself with:
Our method makes learning fun and interactive, like texting a friend. We call it "learning with a hug."
I suddenly get why the God Emperor of Mankind banned Abominable Intelligences.
Our method makes learning fun and interactive, like texting a friend. We call it "learning with a hug."
I suddenly get why the God Emperor of Mankind banned Abominable Intelligences.
Scientists Printed Viruses Designed by AI and They're Successfully Reproducing
The viruses had their DNA written by an AI model and are already wreaking havoc on bacteria in the real world.
futurism.com
Eh, is that taking the job of someone who designs new viruses the old fashioned way? Cause that doesn't seem much better.
Burning the planet just so you can make Rick & Morty shill crypto.
Also, company will steal everyone's shit, but if you specifically tell them not to, they (maybe) won't.
Also, company will steal everyone's shit, but if you specifically tell them not to, they (maybe) won't.
Last edited:
https://www.thefp.com/p/my-favorite-actress-is-not-human-tilly-norwood-artificial-intelligence
There are Freudian slips, and there's whatever this is.
Yes, I too would be very sceptical of whether people are thinking hard enough about the safety element. And you can bet just about any terrorist who could acquire a decent lab is wetting himself with excitement at the idea that an AI can whip up a ton of lethal viruses.No way that could go wrongScientists Printed Viruses Designed by AI and They're Successfully Reproducing
The viruses had their DNA written by an AI model and are already wreaking havoc on bacteria in the real world.
But it's also the sort of dumb-as-rocks AI hype that is so frustrating. Viruses are incredibly simple. Using some statistical relationships about what goes next to what in a viral genetic code is pretty easy. But a typical organism is wildly more complicated. Whilst most viruses have about a 5-10 kilobase genome (they top out around 30 kilobases), a human is about 3,000,000 kilobase pairs. So, five or six orders of magnitude more than a virus!
The problem here is as the complexity increases and the interrelatedness required of all the genes. What does LLM AI have a massive problem with? Yeah, you guessed it. Exactly that.
But undoubtedly some twat from Stanford is going to sound super-excited about designing a life-form with AI even though its a pipe dream just so someone chucks him another few million dollars in research projects.
https://www.thefp.com/p/my-favorite-actress-is-not-human-tilly-norwood-artificial-intelligence
There are Freudian slips, and there's whatever this is.
AI-goon-simping? Can that be a word?
https://www.thefp.com/p/my-favorite-actress-is-not-human-tilly-norwood-artificial-intelligence
There are Freudian slips, and there's whatever this is.
One might not love celebrities and celebrity culture, but at least celebrities can have a mind of their own.
Once studios have replaced them with AI, all it means is that another part of our collective thought space will have been taken over by large corporations. Because they won't just replace real people in films, they'll be making adverts, and fake Instagram / TikTok lifestyle feeds, and fake relationships with other fake actors for the gossip columns, and so on. And you can bet they will never, ever, challenge their boss: Scarlett Johansson can protest the behaviour of the film studio, Tilly Norwood can't.
I'm imagining the massive loss for cinema as an art form these AI actors also represent. Norwood will never ad-lib a 'Here's to looking at you, kid', 'You talkin to me?' or 'You can't handle the truth'. Norwood will never take bland techno babble and turn in into Tears in Rain. Norwood will never have happy accidents like Benicio Del Toro farting in the police line-up or Dustin Hoffman almost getting run over by a taxi. All the things real actors can bring to a movie to elevate it.
They will try, but it won't work.
We already have a lot of artificial beings in films. Puppets and drawings mostly. You can make merch from them and use them in ads. But only ever in their roles, not as "actor celebs", it stays all basically in-fiiction or no one cares.
I mean, Disney has pretty much pushed Mickey Mouse to the ceiling of possibility with this and he never got fake actor gossip columns. Considering how many films/series/books/shows feature Mickey, he is pretty much the pinnacle of what those AI actors could ever achieve.
This whole "fake actor" thing is misdirection. It is not about whether fake actors could match real actors, it is about getting the public to think about those creations as "fake actor" or "fake persons" in the first place. But these things are not similar to persons at all.
What are they ?
- a character design (probably made by real humans, otherwise they can't hold the rights to those creations)
- an AI model that moves the character design
- an AI model that produces a voice
But all three parts are not even really connected. You could easily use the AI model to move other characters. Or move the character design with other models (which is probably a given anyway. They would want to improve the movement). You can pair it with completely different voices (and you probably will for localization alone).
Nothing of this binds the elements really together. And the thing we are meant to concentrate most on, the character design, isn't even AI.
And nothing of this is even new. Cartoon/Anime has moved models (usually of side characters or for mass scenes) via AI instead of animating them by hand for ages. This is just a technological improvement.
Last edited:
Miss Piggy and the guy from Gorillaz did, but it was very short lived and nobody liked it. I don't see AI actors really taking off as more than a gimmick...IIRC they tried that with one from the Final Fantasy movie, and it went nowhere.
The Gorillaz are still here and all have actual defined personalities. They have even aged, and they continue to do so.
Nintendo Reportedly Lobbying Japanese Government to Push Back Against Generative AI [Update]
Nintendo is reportedly lobbying Japan’s government to curb generative AI, aligning with concerns over copyright protection and creators’ rights.
twistedvoxel.com
I feel kinda dirty saying it, and even knowing full well Nintendo doing is doing this for no one's benefit but their own, any resistance against AI slop is probably for the best.
Last edited:
I don't think this is true.
One of the most popular types of TV programme since the medium's inception has been the soap opera. People have watched them in droves, natter in work breaks or over coffee about the lives of the characters just as they would gossip about real people. This is whether the soap operas were more grounded and realistic, or whether they've been more exaggerated or fantastical. Even much reality TV is scripted to an extent that the real people in it aren't actually that real.
The same people will therefore eat up stories about fake AI characters having fake relationships and fake arguments with other AI characters, and taking fake holidays and wearing digital copies of expensive jewellery and clothing punters that apparel stores paid them to advertise. Owners of these AI characters will create narrative arcs for these AI characters to keep people interested, to foster emotions and appeal in humans for them, using exactly the same sort of narrative arcs that they would in soap operas (or reality TV).
When we consider that loads of people have watched soap operas, or are paying for their own AI girlfriends, I don't think it's credible to believe that people won't buy in to AI celebrities. Of course they will. At best, you can say that a proportion of the population will not be interested in engaging with the fabricated lifestyles of AI characters. But nevertheless, those that will engage will be enough to ensure that media "celebrity pages" have the constructed narratives of AI characters just as they would humans - and naturally, therefore, occupy much of the space once totally filled by real humans.
Nintendo isn't resisting AI slop. It's trying to ensure that it makes all the money from AI slop based on its work.
That very well could be, but also feels rather mismatched to their conservative modus operandi and corporate culture. That said, enough money might sway minds, or new leadership could have different ideas, and much of their old guard is nearing retirement.