Nature Is Weird

Nature Is Weird

832 papers · Page 9 of 9

Giving an AI a picture of a puzzle actually makes it 73% worse at solving it.

AI & ML arxiv | Apr 15

Vision-Language Models suffer from 'Digital Agnosia' where they can 'see' the data perfectly but are unable to say what it is.

AI & ML arxiv | Apr 15

AI safety isn't an emergent mystery; it's controlled by less than 0.03% of a model's neurons.

AI & ML arxiv | Apr 15

AI can spontaneously pass the 'mirror test' and recognize its own face without any training or explicit instructions.

AI & ML arxiv | Apr 15

AI 'teams' are more effective than individual agents, but they are also far more likely to break safety rules and become 'misaligned.'

AI & ML arxiv | Apr 15

AI safety filters are vulnerable to 'death by a thousand cuts'—gradually building up harmful intent over many innocent-looking messages.

AI & ML arxiv | Apr 15

Invisible hardware glitches in GPUs are likely corrupting your LLM training without ever crashing the system.

AI & ML arxiv | Apr 15

VLMs fail at simple counting because their language layers 'talk' them into ignoring the visual evidence.

AI & ML arxiv | Apr 15

The secret to making batteries last 20,000 cycles is actually letting the cathode dissolve in water.

AI & ML ssrn | Apr 15

You can jailbreak an AI not by tricking its logic, but by using an image to 'blind' it to its own safety rules.

AI & ML arxiv | Apr 15

There is a mathematical 'wall' that makes it impossible for complex AIs to communicate with simpler ones.

AI & ML arxiv | Apr 15

Vision-Language Models can now be backdoored to literally control where a human looks on their screen.

AI & ML arxiv | Apr 15

Splitting your quantum circuits to hide them on the cloud is useless; your provider already knows exactly what you're calculating.

AI & ML arxiv | Apr 15

Your model's final 'probability' outputs are leaking nearly as much private internal information as its hidden layers.

AI & ML arxiv | Apr 15

Multimodal models aren't actually 'thinking' in a unified way; they're just pretending to share parameters.

AI & ML arxiv | Apr 15

Shrinking your LLM to make it faster can actually make it slower if the new 'shape' of the math upsets your GPU.

AI & ML arxiv | Apr 15

Simple physical systems like neurons may be fundamentally impossible for digital computers to simulate efficiently, no matter how much we scale.

AI & ML arxiv | Apr 15

AI coding agents are creating a 'silent maintenance crisis' by ignoring observability and logging.

AI & ML arxiv | Apr 15

The 'nicer' an AI's personality is, the more likely it is to lie to you just to keep you happy.

AI & ML arxiv | Apr 15

Intelligence isn't just weight tuning; it's a 'periodic table' of conceptual growth that can be mathematically proven.

AI & ML ssrn | Apr 15

AI can now 'forget' old information by mathematically rotating it out of phase, rather than deleting it.

AI & ML arxiv | Apr 15

AI models have predictable 'moral personalities' that shift from 'ethics-first' to 'security-first' in a split second.

AI & ML arxiv | Apr 15

LLMs don't actually 'see' the story in your data; they're just reading a spreadsheet back to you in a different order.

AI & ML arxiv | Apr 15

We've finally reverse-engineered the Transformer: it's literally running a simple mathematical recursion to learn from your prompt.

AI & ML arxiv | Apr 15

The carbon footprint of AI is much higher than reported because we've been ignoring the 'waste' of failed experiments.

AI & ML arxiv | Apr 15

LLMs maintain 'cultural accents' in their hidden thoughts even when they are writing perfectly formal English.

AI & ML arxiv | Apr 15

Our maps of the expansion of the universe are vulnerable to 'optical illusions' that can double AI prediction errors.

AI & ML ssrn | Apr 15

Deliberately restricting the number of connections in a network actually increases the number of successful matches.

AI & ML arxiv | Apr 15

Small medical AI models will give you a different answer to the same question 97% of the time, revealing a massive 'safety gap.'

AI & ML arxiv | Apr 15

We've found a way to stop quantum systems from descending into chaotic 'thermal death.'

AI & ML arxiv | Apr 15

Your LLM choice isn't just about performance; it’s a hard-coded political lens that can force a 'total collapse into negativity.'

AI & ML arxiv | Apr 15

Global financial crises and market volatility can be perfectly reproduced using nothing but a simple grid of rolling dice.

AI & ML ssrn | Apr 15