Captcha tests tell us something terrifying about the future
Have you noticed Captchas have changed?
They’re becoming a bit weird, and harder, as the ‘I am not a robot’ prompt becomes easier for AI bots to solve.
As a result, people have noticed the strange puzzles developers are putting forward to challenge AI capabilities.
In his most recent Netflix special, comedian Jack Whitehall even asked the audience: ‘Is it just me, or have those “I am not a robot” tests started getting harder?’
New versions of the Captcha tests ask users to select options that are the same shape, to move a pointed hand to face the same direction as an object, or to click on the non-aquatic animals.
Captcha – an acronym for ‘Completely Automated Public Turing Test to Tell Computers and Humans Apart’ – was created in 2000 to prevent bots from attacking networks and websites.
They have been used for more than two decades as a way to block automated systems like bots from performing certain actions on a website, which limits spam and stops non-human users from accessing specific resources or web pages.
However, researchers at the University of California Irvine found that bots could consistently answer Captchas’ distorted text with almost 100% accuracy.
The study authors wrote: ‘Advances in computer vision and machine learning have dramatically increased the ability of bots to recognise distorted texts and by 2019, automated tools achieved over 99% accuracy.
‘Alternatively, bots often outsource solving to Captcha farms – sweatshop-like operations where humans are paid to solve Captchas.
‘In light of this, Captchas have changed and evolved significantly over the years.’
Despite the puzzles being created to stop the bots, the fact bots can solve more Captchas and their ever-increasing level of difficulty could mean the difference between getting tickets to a show or standing outside the concert hall, looking for a scalper.
‘Things are going to get even stranger, to be honest, because now you have to do something nonsensical,’ Kevin Gosschalk, the founder and CEO of Arkose Labs, a web security firm that designs captchas, told The Wall Street Journal.
‘Otherwise, large multimodal models will be able to understand,’ he added.
And things have been getting weirder.
One person took to X to ask for help as they’re becoming a bit more complicated: ‘WTF!? @RoyalMailHelp your Captchas are getting so complicated not even humans can figure them out!
Another person took to X to complain about the qualities of the Captchas, saying: ‘ It hurts my brain to look at something that looks realistic but incorrect.’
A third person wrote: ‘One day… I might not be smart enough to pass this “I am not a robot” test.’