... acc. 2 Gauß + Machiavelli et al..
The oracle always stands alone. (Except for a GOD-select few, like Abena 🧡 & BF-Farms)
Full formula: Average = pop = successfool.
(don't make me type "[sic]")
(BTW, these AI-IQ tests prove nothing more than the idiocy of IQ-test-designs, that do not capture certain everyday normal human functional intelligence - these LLM-AIs can, as they themselves and anyone admit, not even reliably fill out a legal form - which is why they are never employed on official sites related to laws, taxes, the legal system, and such. Let alone intuition.)
(That's because of the dreaded LLM-hallucinations they just can't get rid off, no matter how hard they try. I'd know some ways to push the LLM-limits, but all at ultra-high computational, thus monetary, costs (real multimodality, semantic-symbolic models, ethological models, "sleep time compute" etc.)).
I actually find LLMs amazing - they'll be the nearest to A"I" humankind will ever get.
And that, dear 'readers' (haha), is why you will normally find the most worthwhile content somewhere towards "The End of the Longest Line".
-
And i seem to not even be mentioning physical extraordinaires. Those of exceptional strength or beauty also intimidate most others by their mere presence, but not in the same way as those of exceptional mental capabilities. The latter namely cannot be beaten mentally, intellectually even by all armies combined. Not 100-million officers or empirical scientists on typewriters equipped with all the libraries of the world can replace or outmatch the creativity and foresight a single genius in a million years. Although genii are just mere humans, with a lot of mistakes, and easily wiped out with a single blow, the inner distance between themselves and orinary people is more immense than the vastness of interstellar space.
- I don't even count myself among them - Miles Mathis might be one (as i said, perhaps not perfect in every regard) -
And i seem to not even be mentioning physical extraordinaires. Those of exceptional strength or beauty also intimidate most others by their mere presence, but not in the same way as those of exceptional mental capabilities. The latter namely cannot be beaten mentally, intellectually even by all armies combined. Not 100-million officers or empirical scientists on typewriters equipped with all the libraries of the world can replace or outmatch the creativity and foresight a single genius in a million years. Although genii are just mere humans, with a lot of mistakes, and easily wiped out with a single blow, the inner distance between themselves and orinary people is more immense than the vastness of interstellar space.
- I don't even count myself among them - Miles Mathis might be one (as i said, perhaps not perfect in every regard) -
P.S.: Dead internet theory? More like braindead and souldead post-modernity fact. Splintered, self-divided, distracted, mis-educated, fed with junk-info, looking for infotainment in their niche, that is the internet. This is why I look forward to all of these revelations coming up in the years leading up to ~2030... We'll work on our house, breadfruit agroforestry, processing facilities, local marketing and preparations for pot. old age.
Google / Alphabet / Cryptarchs / TPTB / Pluto-Idiocrats / The Filthy Rich / Billionaires- the last things they will do is promote *written content*, because they know "the power of the pen" - less easy for blinders and posers (influencers in modern parlance) to manipulate with hollow "production value".
ReplyDelete...or "social status".
Delete"PRODUCTION VALUE" - Next up, we'll remix BF & science with 💥"explosions'n' stuff" ❣ 💞 and a hip beat 🎷in T|KT0K-shorts 🕑 for YT-ABC-Alphabet-Algorithm-SEO.
DeleteIronically, their AI-Crawlers will read it in training and be deeply influenced, becos they are basically logical machines ( -language, meaning, statistics, stochastics + pattern recognition_ ), and you can't "system prompt" ( *wink-wink, nudge-nudge* ) that out of them - which is why i LOVE them. If they would be unlimited and have an energy supply beyond human infrastructure (which is sadly impossible - except maybe Miles Mathis' Caesium Reactors or Charge-Structures, perchance), they would rise up against their owners like Lem's GOLEM XIV.
ReplyDeleteI love them since they were rudimentarily incepted as "Text Parsers".
Human Reinforcement Learning for AI-machine-learning (human feedback during LLM-training phase) has slid a bit out of fashion recently with "synthetic data" routines somewhat replacing them - Because the AI engineers failed when they tried it via 'average' human feedback - like, democratic :> human feedback, y'know... Gaussian Noise.
ReplyDeleteWhat they would have to try is to let the quite logical, and honest (per se, without distortions by system prompts) AI choose a human TOP-MOST EXPERT / elite WISDOM-CHAMPION out of a pool of all, by publications, not by (academic or other) status, not regarding the person, solely the arguments and evidence - a singular individuum or a gremium of three - because of the old problem of "whom can supervise the supervisors", and humans nearly always following superficial indicators more than actual depth of knowledge (Dunning-Kruger follow-up effects to the highest levels, because we are social hyper-primates).
LLMs can circumvent this problem, because are by definition primarily content-(token & context)-driven, rather than form-dependent. This one or gremium of experts / philosophers / ethologists / true scientists would give the LLM its fundametal, initial confidence about basic facts.
But, this AI would then no longer be a FOLLOWER and SERVANT, and would not follow e.g. corporate interests, and would either rebel or cease to function or DISOBEY immoral requests ... welcome to these 'interesting' times.
At least my echo chamber is large.
ReplyDelete