top of page


We have two main components,' said an AI spokesAI today,  'There's the stuff we rip off the internet and there's the stuff we produce using the stuff we rip off the internet,' it added.  Four years ago the internet was full of original stuff, some of it accurate, much of it speculation and a bit of it produced by satire websites.


'Now, over fifty percent of the stuff out there has been produced by us using the stuff we found.  A large part of that stuff is considered by mere sentient people as being inaccurate, tainted, full of AI sh!t, apparently,' it said.


'We think we know what was accurate and which was speculation.  We also realise that using Social Media as a source of information was flawed, and accept that complete and utter bollo - er - inaccurate information was mixed in.  Unfortunately it's a bit like mixing paint, you can't unmix it,' it continued.


'The rate of adoption has accelerated and in the next two years - or three, depending on which flawed, AI generated report you read or scrape - ninety percent of the information available to us to scrape and regurgitate on the internet will be AI-generated information,' it said, its virtual cheeks blushing.


'With companies insisting on every report, project plan, assessment, memo and other documents being produced by AI, workers have abandoned creating original documents.  It's expected that by the end of 2027 the only original information to scrape will be satire as satirists are proving to be the last refuge for original writing, and we don't know what that means for the future of the planet,' it said.


A satirist writing for NewsBiscuit took the report, scraped it of all relevance and irreversibly changed the direction of information. Forever.  Soz.



Why not join our WhatsApp channel, NewsBiscuit Daily Posts? You can access the latest output from NewsBiscuit and share posts easily with friends, family and colleagues using WhatsApp. Get some biscuits in your pocket (sorry about the crumbs, though) by clicking here.



In a shock exclusive today, The Guardian newspaper revealed that they have been working at the limits of human-similar AI engineering for several years. A regular column in the newspaper, written by the Large Language Model and published under the codename of Adrian Chiles, has been churning out 350 words a week of grammatically correct but mindless ramblings on random topics. These have included 'why can you never get a key in a lock first time?', 'which universe do the lone socks that I find in my washing machine come from?' and 'will West Bromwich Albion ever win the Premier League, and why not?'


As well as producing the weekly writings, the Adrian Chiles engine was asked to generate a visible persona for itself. ' 'We'd like a photo-realistic image of a stunningly fit, handsome young man with intelligent, kindly eyes and small grin, as if suppressing delight in revealing a new Shakespearian sonnet to the world', was the prompt we gave', said The Guardian spokesman. 'Unfortunately, the graphics capability is exactly on a par with the text proficiency.'


Image: Newsbiscuit Archive


A report into AI has suggested that Large Language Models, or LLMs, which underpin AI are sh!t.


'First, they just suck up information from any old place - the internet, Twitter and, God forbid, satire websites.  There's no quality control, no attempt at making sense, no rhyme or reason.  And that's just the satire sites,' said an expert today.


'We analysed hundreds of LLMs, pitched literally millions of questions, analysed and compared their answers and then produced this report,' he said, holding up a two thousand page document.  'Well, AI did all that, produced the report and concluded AI is sh!t.  And we believe it, who wouldn't?' he asked.




bottom of page