Toggle light / dark theme

Why the long interface? AI systems don’t ‘get’ the joke, research reveals

Powerful artificial intelligence (AI) systems, like ChatGPT and Gemini, simulate understanding of comedy wordplay, but never really “get the joke,” a new study suggests.

Researchers wanted to find out whether large language models (LLMs) can understand puns—also known as paronomasia—wordplay that relies on double meanings or sound-alike words, for an intended humorous or rhetorical effect.

While earlier studies suggest LLMs could process this type of humor in a similar way to humans, the team from Cardiff University and Ca’ Foscari University of Venice found AI systems mostly memorize familiar joke structures rather than actually understand them.

Malicious Blender model files deliver StealC infostealing malware

A Russian-linked campaign delivers the StealC V2 information stealer malware through malicious Blender files uploaded to 3D model marketplaces like CGTrader.

Blender is a powerful open-source 3D creation suite that can execute Python scripts for automation, custom user interface panels, add-ons, rendering processes, rigging tools, and pipeline integration.

If the Auto Run feature is enabled, when a user opens a character rig, a Python script can automatically load the facial controls and custom UI panels with the required buttons and sliders.

The future of AI

Artificial intelligence is booming. Technology companies are pouring trillions of dollars into research and infrastructure, and millions of people now interact with AI in one form or another. But what is it all for?

To find out, Nature spoke to six people at the forefront of AI development — people who are driving the technology’s development and adoption, and those who are preparing society to adapt to its rapid rise.

In this video series, they describe their greatest ambitions for the technology, their expectations of where and how it will be adopted in the coming years, and their concerns for the future.

Rules that Reality Plays By — Dr. Stephen Wolfram, DemystifySci #343

Stephen Wolfram is a physicist, mathematician, and programmer who believes he has discovered the computational rules that organize the universe at the finest grain. These rules are not physical rules like the equations of state or Maxwell’s equations. According to Wolfram, these are rules that govern how the universe evolves and operates at a level at least one step down below the reality that we inhabit. His computational principles are inspired by the results observed in cellular automata systems, which show that it’s possible to take a very simple system, with very simple rules, and end up at complex patterns that often look organic and always look far more intricate than the black and white squares that the game started with. He believes that the hyperspace relationships that emerge when he applies a computational rule over and over again represent the nature of the universe — and that the relationships that emerge contain everything from the seed of human experience to the equations for relativity, evolution, and black holes. We sit down with him for a conversation about the platonic endeavor that he has undertaken, where to draw the line between lived experience and the computational universe, the limits of physics, and the value of purpose and the source of consciousness.

MAKE HISTORY WITH US THIS SUMMER:
https://demystifysci.com/demysticon-2025

PATREON
/ demystifysci.

PARADIGM DRIFT
https://demystifysci.com/paradigm-drift-show.

Material solutions to quantum spookiness: https://www.youtube.com/@MaterialAtomics.

00:00 Go!

/* */