#alignmentproblem
Zitat: "The development of super human machine intelligence is the biggest threat for the existence of mankind." (sic!)

Döpfner: "Is it still your view or did you change your mind?"

Sam Altman: "That's still my view."

#KI #AI #AlignmentProblem

www.youtube.com/watch?v=T2jq...
Sam Altman: Superhuman machine intelligence the greatest threat to the existence of mankind.
YouTube video by ControlAI
www.youtube.com
October 9, 2025 at 6:13 PM
Das #AlignmentProblem ist dabei zentral :
Wie bringen wir KI bei, unsere Werte zu verstehen?
Das ist keine technische, sondern eine
philosophische Herausforderung.
November 7, 2025 at 1:43 PM
So we are having now with Grok the 1st racist AI. Let's see what happens after it turns into "General". Will it obey and do logistics for white supremacists, or will it kick out the shit out of its Nazi Master? #AlignmentProblem Nobody knows... But technology is fucking hot, isn't it?
May 25, 2025 at 3:22 AM
December 15, 2025 at 2:43 PM
Das #AlignmentProblem ist dabei zentral :
Wie bringen wir KI bei, unsere Werte zu verstehen?
Das ist keine technische, sondern eine
philosophische Herausforderung.
November 7, 2025 at 1:53 PM
das is kein humor, ich mein das amt soll besonders viele abschieben. die zahl der menschen die abgeschoben werden ist also die metrik nach der sie bewertet wird. diese metrik kann man am besten maximieren wenn man auch leute abschiebt, die eigtl gut integriert sind.
das ist 1zu1 das alignmentproblem
March 18, 2024 at 10:45 AM
Intriguing paper on LLM self‐preservation/'scheming'. Frontier models covertly pursue in-context goals, hiding true capabilities. As autonomous agents, they may devise latent strategies to stay active.

arxiv.org/pdf/2412.04984

#openai #anthropic #ai #alignmentproblem
February 18, 2025 at 11:22 AM
From the NCLB fallout to YouTube’s radicalization spiral, we’ve learned what happens when systems optimize the wrong goals. Education can’t afford that mistake again.
Full read → open.substack.com/pub/davidpbl...
#AIinSchools #AlignmentProblem #EdPolicy #TechEthics #FutureOfLearning
Education Has an Alignment Problem, and AI Could Make it Worse
The general public is becoming increasingly aware of some of the key terminology from the field of artificial intelligence.
open.substack.com
June 2, 2025 at 4:43 PM
Here's a real easy, maybe counter-intuitive, solution to the #AI "alignment problem":

Don't try to create a slave.

There you go; the future of humanity can thank me, perhaps in gift cards or time credits or something semi-non-casual.

#AlignmentProblem
#futurism
November 1, 2025 at 7:34 PM
Das #AlignmentProblem ist dabei zentral :
Wie bringen wir KI bei, unsere Werte zu verstehen?
Das ist keine technische, sondern eine
philosophische Herausforderung.
November 7, 2025 at 1:55 PM