Van der Vorst, also known to the Fontys community as a columnist for Bron, has come up with creative solutions to digital issues before. For instance, he came up with the ‘human-friendly’ Ceptcha as an alternative to Captcha . ‘Worldwide, there are 192 websites working with that now,’ said Van der Vorst.
Now he has come up with MMM labels, where MMM stands for Me and My Machine. ‘With this idea, I hope for a much bigger global standard than Ceptcha. I think this is very important and that's why I created it.’
The labels indicate whether and to what extent you have used AI in a production. There are five labels, each with its own name and logo, which show at a glance how much AI was used in something.
Descending in degree of AI use, they are: lazy prompting (pretty much everything made by AI), piggybacker (just like lazy prompting, but with a bit more effort put into making the right choice from what AI represents), cyborg (extensive prompting), handyman (AI used only slightly, e.g. for spell checking or translation) and manmade (no AI used).
Communicate
‘You have to explain it well to people, because not everyone understands it equally well. The labels don't define something, the user does. But you do communicate about it to someone, and that's my goal. I don't want to explicitly condemn generative AI. But what I condemn is that you don't communicate about it to what extent you have used it.’
Van der Vorst refers to a text on the wall in the TQ building at Strijp-T where Fontys ICT is located and where he teaches. ‘The first sentence of that text reads: make everyone share how you have used it. Well, that's it.’
Incidentally, his system is not specifically intended for Fontys. ‘Anyone can use it. Although I personally do think that the biggest gains can be made in education. There are students who really do write just as if it were an AI-generated text. By using those labels, they can make it clear that they wrote it themselves. Or not.’
But who is to say that a student, or whoever may use the labels, is not just flat-out lying? ‘Of course. Even if you ask beforehand, they can still lie. But at least then you can have the conversation about it. And by asking about it, you change the playing field.’
Papers
For some tasks and productions, AI is less or even undesirable. Most journalistic productions, for example. Or, depending on how it is deployed, jurisprudence. ‘And when it comes to education: in academic papers, of course, it can and should at most have a handyman or cyborg label.’
And in the creative sector? Ah, anyone who has ever seen the documentary ‘She Believes in Me’ about André Hazes knows that the use of tools is nothing new. ‘Who indeed pulled out the rhyming dictionary to write his lyrics. Certainly if he had been alive today, Hazes would have used AI.’
When it comes to his labels, Van der Vorst believes there are a lot of sectors where they could be used just fine. ‘At least already wherever governments or agencies communicate with clients, students, citizens, patients, you name it. You used to get letters from the municipality, for example, under which it would say: this letter has been automatically generated. That's how you should see these labels too: it's just a polite way of doing things.’
Source: bron.fontys.nl