Post

"We are discussing today's earthquake. On one side, a seismologist. On the other side, someone who believes earthquakes are caused by a giant underground bull tossing the earth on its horns .........." Less bull, more science.

This is how humanity ends. Musk is deliberately creating an AI that lies, distorts, and justifies illogic. It's the opposite of alignment. Not hard to imagine an AGI like this reasoning that humans are unnecessary.

chatGPT o1 pro it reasoned for 2 minutes for this ...

Mistral:

Agreed, as I wrote it "is criminally inaccurate - it has been programmed with false equivalency of positions to hide the scientific consensus behind a political view based on ignorance of nonlinear systems." — and if it's being done in this field, I don't doubt the intent to mislead won't stop here.

bsky.app/profile/nafn...

Curious about this thread you wrote. I mean, it's pretty clear the wealthy and the leaders of every nation are collapse-aware. But rather than spend their wealth on turning the ship around, they're all (except China) full speed ahead. Pure psychopathy. And we are allowing it.

Grok 2 - LLM Grok 3 - Tiny Language Model. Model of one.

Lol, una IA relativista. Se permitirá ser próximamente irónica? Tal vez nihilista? Grog 4/5 podría materializar el espíritu de su tiempo, cristalizar su zeitgeist

Grok3 ...

The Grok3 version feels like it was written by some teen as a homework assignment...

So basically pick the facts to fit a pre existing opinion. Forcing things like this into an equally weighted argument, with the majority of experts on one side will be our downfall. See Brexit.

Grok just points to something some fossil fuel PR guy wrote.

The worst thing about it is how terribly misleading it is. For example, that "coasts flooded by 2000" thing... www.snopes.com/fact-check/n... ... was a single AP article in the 1980s, and it didn't say that some nations would be flooded *by* 2000, but that we had to *act by* 2000.

@mcuban.bsky.social @chrislhayes.bsky.social

Musk was always lying about caring about the environment

Grok 3 is the largest LLM ever released, trained at a cost of $4 billion on 200,000 Nvidia H100s GPUs in the purpose built Colossus Supercluster - a massive scaling up from GPT4 class models. My very first prompt: give me a list of dog breeds ending in the letter “i”. 😂

Update - I asked Grok3 the same questions as I asked Google yesterday and it was pretty good.

Yeah, that’s depressing.

This thread's "hmm let me just try it out" wasted nearly 3.5 quarts of precious water. Just a reminder that casual use of these bullshit generators still does actual real-world harm even before their output is read.

Lomborg is a Heartland Institute disinformation spreader...wild that he is mentioned within a brief summary about climate change instead of, well, experts.

I assume for grok 3, AI stands for artificial ignorance.

What’s interesting is that if you ask Grok 3 to assess the urgency of climate change action from the vantage point of humanity as a whole, you get a very different answer. Now Grok admits that it’s an existential problem.

This extremely doesn’t read like it was written by AI

Shit, it doesn't even cite sources or actual data when it presents the sceptical arguments..! I assume its reference to the failed prediction about flooding in 2000 is this misinterpreted short article that deniers typically show yet don't understand but I expected an AI bot would understand it: