Subscribe now

Technology

Silicon Valley’s top AI models are terrible at rebus wordplay puzzles

Rebus puzzles provide wordplay challenges involving both images and text, and they can confound Silicon Valley’s most powerful AI models

By Jeremy Hsu

25 January 2024

Rebus that works out as Beethoven

A rebus puzzle for the composer Beethoven

Arjun Panickssery

Cutting-edge artificial intelligence models such as OpenAI’s GPT-4V and Google’s Gemini struggle to solve clever wordplay puzzles involving both images and text.

Rebus puzzles typically require a puzzler to identify a word represented by an image, to add or subtract letters from that word and to combine the result with words identified from other images to arrive at a solution.

For instance, a rebus might include a picture of the planet…

Sign up to our weekly newsletter

Receive a weekly dose of discovery in your inbox! We'll also keep you up to date with New Scientist events and special offers.

Sign up

To continue reading, subscribe today with our introductory offers

View introductory offers

No commitment, cancel anytime*

Offer ends 2nd of July 2024.

*Cancel anytime within 14 days of payment to receive a refund on unserved issues.

Inclusive of applicable taxes (VAT)

or

Existing subscribers

Sign in to your account