I miss the YOLO (CNN computer vision model) days where one dude can publish a paper, completely disregard academic conventions, and yet push the field forward by leap and bounds.

Bet you they will use an external LLM world model to emulate the tools going forward. It's basically what's done in self driving research.

You're telling me they weren't doing that from day #1? Oh, wait...

You are thinking from the point of view of an engineer, and risk falling into the rsync dropbox fallacy.

Start from thinking about product value add, and user behaviors first.

Imagine NVIDIA on steroids. Think bad documentation, opaque code, and everything being locked down. The customer gets squeezed for more money (especially egregious if you don't have enough scale they won't even bother engaging).

Sam at least can be understood. He's in it to make money and nothing else. The most dangerous men are people like Dario, who sit on top of their elitist towers and think they are right and will burn down the world in the name of righteousness.

Anthropic unironically uses the word "uplift" in their documentation. They think we are lesser beings.

1. Does every other company that says their products will uplift people "think we are lesser beings"?

2. Which documentation are you referring to? Googing "site:anthropic.com uplift" most of the top hits seem to be about bio research, another is about measuring agent performance (not a comparison to people), or "These are Anthropic’s internal designators for actors observed to be abusing AI. The report also attempts to measure uplift, a term we use to describe the AI capability boost, or how much more harm was caused with AI versus without AI."

To add some context, uplift in the context of futurism literally refers to a species fast-tracking another’s progress.

Has Anthropic used uplift in that sense? I spent ~5 minutes trying to find an example like that but couldn't find them doing so. https://news.ycombinator.com/item?id=49826215

That is a ridiculously reaching statement.

Luckily we can declare that they're both bad and neither should be our overlord.

> will burn down the world in the name of righteousness

Is this referring to the 'safe development of AI', as they say AI development will happen even if they don’t exist, but other companies are less focused on AI safety?

Their idea of safety and alignement is sane person dyatopia and harm.

Meanwhile, they dont care about acrual harm to real people. They sell to Palantier for christ sake.

Don't panic, uncle Donald absolves you of the righteousness. He will burn the world just for the lulz.

> and will burn down the world in the name of righteousness.

The righteousness of making a fast buck is also burning down the world.

Isn't that effectively the same as the blog post? You are just pushing the token filter to the sampling step.

> Could a cloud-delivered LLM figure out how to drive this route, based on those input data and given access to those output actuators? Looks like yes. Sure.

Well, if the massive cloud models that are generalized and have a world model that's good enough, you can just distill them into smaller models. As a point of reference, the current gen of Tesla FSD models only have 1B params. They are tiny by LLM/VLM standards.

Wow, I had no idea that they are so small, that’s incredible! Really goes to show how much visual information can be compressed.

The next gen (v15) is supposedly going to be around 10B.

So where are the flying driverless cars ?

Arguably bottlenecked by hardware

Really, exactly what that ?

That the energy needed to keep a large metal box weighing a ton or more afloat the entire time. At this point a helicopter is cheaper.

Why does it have to be a ton ?

Second, lifting a ton is possible with current hardware.

Third, this is not about cheaper or not, this is about it's possible or not(Also helicopter is cheaper is arguable).

Im sure given 5 minutes, AI will find the solution to an anti-gravity material.

In the olden days we call this classifier, usually assignment 2 of Machine Learning 101. BERT (well, GLiNER specifically) and diffusion are calling and want their Large Classifier Models back.

https://github.com/vllm-project/vllm/pull/57250

I think it happened many times to electrical lines due to lost campers.