227
The Time Bomb Went Off: AI's All-You-Can-Eat Era Just Ended in Real Time
(www.thestateofbrand.com)
this post was submitted on 18 May 2026
227 points (96.3% liked)
Fuck AI
7070 readers
1737 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
AGI "can't" happen like man can't fly or 256 KB is enough memory. Will or can it happen? No one is sure. Will it happen this way, through plugging more compute and power into LLMs, that's a definite no. But it's the money grab right now, and anyone who didn't follow the leaders would have been left behind with nothing.
So you're right for this discussion, but in the broader sense we can't say that for sure. One thing I'm sure about is how LLMs and the profit scramble for it have ruined actual research into the real thing.
Here's my prediction. So-called "AGI" will not happen at any point in the life of anybody alive today (including people who were born while I was typing this message). See, thing is, we can't even define intelligence in any meaningful way that has general agreement among the academic stakeholders (philosophers, neural scientists, cognitive scientists, etc.) and there isn't any plausible line of inquiry that will change this in the pipeline that I can find.
And you think a bunch of techbrodudes are going to successfully make an artificial version of something that can't even be successfully defined?
Yeah. Not gonna happen.
I think the most "foolproof" method of creating AGI would be to copy the structure of the human brain. Yes, it's extremely complex and is unlikely to be the most efficient solution for a given intelligence level, but we know it works.
That's not going to happen in anybody currently alive's lifetime either. The best we can manage right now is the Cedars-Sinai model of a neuron (note the singular!) that approximates the electrical behaviour of a single human neuron. (There are huge biochemical signalling networks in brains that are a large part of brain function as well.)
We are nowhere near close to emulating a single human neuron down to the molecular level, complete with ion channels and all the other raw complexity of the beasts. We've barely begun working out how the biochemistry interacts with and modifies that electrical activity. And this is presupposing that the Orch-OR hypothesis (the one that posits essentially quantum computing in the neuron via microtubules and assorted mechanisms held within those) is wrong. If that's correct, we're even farther behind on emulating a human neuron.
That's A human neuron.
We have about 86 billion or so neurons in the brain.
Oh, and wait! Neurons may not even be the whole picture! It's turning out that we're finding some "thought" happening outside of the neurons in the brain.
Did I mention that this is all the static structures of the brain? As little as we know about those, we know even less about the dynamic interaction of things. Biochemical weighting. Plasticity vs. rigidity. Dendrite pruning and regrowth. We know none of this yet, except that they exist.
Oopsie!