Featured Image Credit: Getty Stock Images
Topics: Artificial intelligence, Jobs, Science, Technology, ChatGPT
Rhianna Benson
Topics: Artificial intelligence, Jobs, Science, Technology, ChatGPT
Tech experts have detailed a list of supposed features that employers and university professors can keep an eye out for if they fear their staff/students are relying too heavily on artificial intelligence.
Use of the likes of ChatGPT have been steadily on the rise in professional circuits in recent years, since the controversial chatbot app was founded back in 2022.
While it may have been developed as a general question and answering tool, users of AI writing tools quickly discovered it could be used to pen entire essays with human-like specifics, which could be amended and extended with the simplest of instructions.
Naturally, this usage progressed onto seeing university and college students using ChatGPT to either plan their educational assignments - or even write the whole essay for them.
Advert
And soon enough, employers and bosses began suspecting that more and more professional cover letters, job application question answers and even CVs were used with the assistance of AI.
As we say, experts in this somewhat dystopian industry have since addressed concerns surrounding authenticity.
And in a new article, spokespeople from Mozilla, a free software community, have lifted the lid on the supposed ways in which professional leaders can tell whether or not a piece of written content was done so 'by hand'.
The tech site firstly touches on the previous existence of an online tool which was, bizarrely, created by the same brains behind ChatGPT, that supposedly assessed written content and sniffed out 'fraudulently-written' sentences.
Advert
It needed 1,000 words of content to prove useful, however.
And Mozilla bosses say, even then, it 'wasn't perfect', as AI-made text that has been even remotely amended by a human was known to regularly fly under the radar.
Notice how we're describing the OpenAI API Key in the past tense, being that the tool was taken offline in 2024 after being left out of an update made the July prior.
With that in mind, data scientist Jesse McCrosky pointed out that there may never been a totally accurate means of picking out AI phrases.
Advert
"Detector tools will always be imperfect, which makes them nearly useless for most applications," he told Mozilla.
"One can not accuse a student of using AI to write their essay based on the output of a detector tool that you know has a 10% chance of giving a false positive."
Despite the scientists' assertion, this hasn't stopped others, especially members of their Gen Z community, to voice their 'ChatGPT hyphen' theory, claiming it can be used to vet whether bots had intervened in a piece of written.
Advert
As Tyla reported yesterday, a hyphen - also known as a dash - is a piece of punctuation used to break up a sentence, and is common in magazines and news articles.
The symbol can be used as an alternative to a common, a colon or brackets, being that they supply an additional piece of information.
Recently, however, social media experts have issued a warning against legitimate writers relying too heavily on hyphens in their work, arguing they're commonly used in AI pieces of work.
Advert
More specifically, LuxeGen Podcast hosts Daisy Reed and Sapna Rao emphasised recently that chatbots often use elongated dashes without spaces—like this, as opposed to shorter, authentic ones - like this.
The pair specifically called out a popular fashion brand for including a stream of dashes while announcing their recent rebrand, revealing that the brand's top Instagram comment was 'Including the ChatGPT hyphen is insane'.
Issuing a 'public service announcement', Reed continued: "If you’re at school and you’re using [ChatGPT] for your essays, take out the hyphens. I can always tell."
The podcaster went on to encouraged her followers, whether students or professional workers, to check back through any work they've 'written' with the assistance of artificial intelligence.
She advises, instead, putting chucks of text back into their own work to avoid aforementioned detection.
"Person, ChatGPT, person," Rao added.