Blog

CAPTCHAs are hard for computers to solve, but easy for humans

Image: ©iStock.com/Oleksandr Hruts

Double Helix magazine is looking for your questions! Our Microscope column answers the most intriguing science, tech, engineering and maths queries you can throw at us.

Comment down below with your question, or email us at Helix.Editor@csiro.au. The best questions will be published in our magazine! Here’s a sample question to get you thinking.

Casper, age 9, asks
How do ‘I am not a robot’ quizzes work?

If you’ve provided your name and details on a website, you might be asked to click on a box to prove you’re not a ‘robot’. Of course, it’s not asking if you’re made of metal and circuits – it’s asking you to prove you’re not a program.

Some programs are designed to quickly create numerous fake profiles for social media. Others will swamp a website to overload it, forcing it to shut down.

Protective programs called ‘Completely Automated Public Turing tests to tell Computers and Humans Apart’ – or CAPTCHAs – ask users to do something we know computers can’t yet do very well. In some cases, this involves reading looping letters, or choosing photographs with a specific item in them, such as a car or bridge.

But computers are getting better at doing these things. A few years ago, Google came up with something called reCAPTCHA, which requires you to just click a box.

By comparing your computer’s address with things like your browsing history and how you move your mouse, reCAPTCHA can judge whether a human clicked the box or a program checked it!