Skip to main content

You can try out this AI-created magic card trick at home

card trick
Dmytro Surkov/123RF
Writer Arthur C. Clarke famously said that, “Any sufficiently advanced technology is indistinguishable from magic.” With one of their latest research projects, researchers from Queen Mary University of London are carrying out a variation on this idea — by using technology (and, more specifically, artificial intelligence) to create a “mind-reading” card trick.

The trick involves a magician with two decks of cards. These get shuffled, and then the subject is asked to pick a total of eight, including both words and image cards. They must then choose a word and an image, without showing it to the magician. The magic bit? That the magician has already written down the pairing the person has chosen. It seems like spooky stuff, although its creator points out it is basic psychology.

Recommended Videos

“The AI system [that created the cards looked] for words and pictures that are closest in some type of meaning, a measure called semantic similarity,” Peter McOwan, professor of computer science, told Digital Trends. “The algorithm finds sets of themed target words and pictures that match — such as ‘hunger’ — and other sets of words and picture that don’t match as well, such as ‘fashion.’ We then produce printed cards with these sets of similar and different word and pictures, and use a bit of clever maths in the card shuffle and dealing that means each pile of cards dealt has only one target word or target picture in it. When a spectator picks a word packet and a picture packet, it’s most natural to make the word and picture association we want them to, so our prediction proves correct.”

Image used with permission by copyright holder

It’s neat because — at its root — it is a classic trick. There is no EEG brainwave reading, or fancy levitation. Instead, it is using pattern recognition and internet searches to discover how the  mind subconsciously matches words and pictures. It is based on an idea that a  magician could probably do but computers and big data make significantly easier. The researchers successfully tested the finished magic trick on attendees at science fairs — and used this to verify that it worked correctly.

Please enable Javascript to view this content

A paper describing the work was recently published in the journal PLOS One. If you want to have a go for yourself, you can also download the assets here.

“It’s a new area of study, which is always exciting to start to explore,” McOwan said. “Magic has been a hobby for years, but it’s also an interesting way to explore areas like machine creativity and human perception and cognition.”

Next up, he wants to use AI to come up with magic tricks from scratch. Just so long as they do not all involve sawing us pesky humans in half, we’re happy!

Luke Dormehl
Former Digital Trends Contributor
I'm a UK-based tech writer covering Cool Tech at Digital Trends. I've also written for Fast Company, Wired, the Guardian…
Google’s AI detection tool is now available for anyone to try
Gemini running on the Google Pixel 9 Pro Fold.

Google announced via a post on X (formerly Twitter) on Wednesday that SynthID is now available to anybody who wants to try it. The authentication system for AI-generated content embeds imperceptible watermarks into generated images, video, and text, enabling users to verify whether a piece of content was made by humans or machines.

“We’re open-sourcing our SynthID Text watermarking tool,” the company wrote. “Available freely to developers and businesses, it will help them identify their AI-generated content.”

Read more
AI-controlled robots can be jailbroken, and the results could be disastrous
The Figure 02 robot looking at its own hand

Researchers at Penn Engineering have reportedly uncovered previously unidentified security vulnerabilities in a number of AI-governed robotic platforms.

"Our work shows that, at this moment, large language models are just not safe enough when integrated with the physical world," George Pappas, UPS Foundation Professor of Transportation in Electrical and Systems Engineering, said in a statement.

Read more
You can now generate AI videos right in Premiere Pro
An example of the Firefly video generator

Firefly can now generate videos from image and text prompts, as well as extend existing clips, Adobe announced on Monday. The new feature is currently rolling out to Premiere Pro subscribers.

The video generation feature makes its debut in a number of new tools for Premiere Pro and the Firefly web app. PP's Generative Extend, for example, can tack on up to two seconds of added AI footage to either the beginning or ending of a clip, as well as make mid-shot adjustments to the camera position, tracking, and even the shot subjects themselves.

Read more