And if we really want more human-like LLMs, research can and has been providing for that too

I prefer the headcanon that "James Bond" is a codename for a title

I was surprised to find out that solar panels are an order of magnitude more efficient at energy conversion than plants

Follow-up question: where the f are these gradients coming from

Ablations:

- Using just the nonpolar solvent provided some improvement, but not as much as the whole grease mixture
- Using just the fatty acid salt did not provide improvement
- Other amphiphilic substances, such as methanol, did not provide as much improvement as soap in the mixture

"Please buy this box to see the mystery product inside"

Many founding companies' origin stories were that they stuck together assemblages of bargain-bin hardware until they could move to enterprise quality hardware

Giant AI rig consisting of masses of questionable quality rtx '90-tier GPUs

(still waiting for other people to do replication and ablations)

There's this paper, which proposes an architecture that is both neuromorphic and at least on paper competitive with similar transformer architectures

arxiv.org/abs/2509.26507

Though I regret to inform you that unfortunately, this horror is well within my comprehension

Alright, what regulatory change happened here

You can start pressing your genitals _after_ you get the critical parts of your setup working

If you cannot mentally distinguish between, say, 1) Some guy running a local image model on his PC for funsies, and 2) Elon Musk building a datacenter in the most batshit-irresponsible way possible, then you have put yourself in a position where you DO NOT ENCOURAGE responsible or ethical behavior.

I've been online long enough to know that a lot of people just like to be gigantic assholes the moment there's no social pressure to do otherwise, even to people who are doing nothing wrong at all. It's rare to see someone admit it like this. Usually they maintain the facade a bit longer.

The one trap set here is by the person pretending these are the concerns they have with AI, when they neither care if anyone is trying to solve them, nor is that even the actual issue for them, as they'd admittedly be a judgmental shit about it even with zero ethical reason.
bsky.app/profile/plai...

This sounds like a nasty surprise waiting to happen over and over across multitudes of locations