#aidoom — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #aidoom, aggregated by home.social.
-
I really want him to be right. Also I really don't want my parent's pension to collapse...
-
> IMO, you could get the same or very close results with a traditional/non-llm model.
Still the #aislop got two trillion tulip bulbs and a third of world's electrical power production. Because of the #aislop bubble run specialized models, in sec-op and medicine especially, got abandoned and defunded.
> you have to have the same level of knowledge
What I said previously: you need the experience and expertise to use the tools. I just countered the mood of attacking any user and any use of the llm #aidoom machines. There are areas where there is no choice but to check what an adversary might also have checked. I can and I do condemn any other use.
@mttaggart @womble @ferrix @TindrasGrove @Muddobbers @bryce @malwaretech
-
-
Wow, I hardly ever 1) finish reading one of Pluralistic's essays (something about his typographical choices antagonizes my eyes), let alone 2) post a link to it, but this one? This one's a banger: https://pluralistic.net/2026/04/16/pascals-wager/#doomer-challenge
-
RE: https://infosec.exchange/@AmmarSpaces/116408544329003114
I forgot to note, maybe just in here.
This doesn't mean that the model is just a meh, but it is indeed a fact that there are too many stupid misconfigs, human errors that are floating across our infrastructure.
If Mythos were to be released (well, even before Mythos lol), the first victim is not the fancy big tech, it is the small orgs and maybe goverment institutions because security to them is just a hassle.
The doom is because we neglect too much on obvious problems, not because of the fancy "zero-days" that media and Claude advertise as root cause.
#cybersecurity #ArtificialIntelligence #AIdoom #Mythos #anthropic #AI
-
🚨 AI Safety Advocate Linked to Multiple Murders
｢ Obsessed with the concept of "Roko's Basilisk," a ghoulish thought experiment imagining a future artificial superintelligence torturing its opponents for all eternity, LaSota squarely fell into the category of those who think AI will destroy the world — and thought it was her duty to stop it ｣
-
I like these early samples of "voicemail spam by AI" that I'm hearing.
"We've (heavy sigh) found the problem with your Google listing--"
Gee, it sure sounds like it took a lot of work.
And spec work, at that...!
I imagine this method will eventually A/B test its way to the sound of a family member's voice...
...frantically screaming at me to upgrade my Google listing before the ominous mist reaches my front door... -
@Herr_Irrtum leidet das Handwerk wenn die Werkzeuge immer besser werden?.. denke was man kreieren will, mit der Botschaft und den Gefühlen dahinter ist da essenzieller. Was die Ki da abliefert ist per Definition durchschnitt(einfach ausgedrückt).
-
#AiDoom #ai #ki #kreativ #Musik
KI in #Logic DAW.
Als jemand, der all das selber hinkriegt und Spaß an diesem Prozess hat:
Ich find's maximal zum Ko… zum Übergeben!
Eine Kultur des Misstrauens und der Nicht-Wertschätzung von Kreativität steht nicht nur bevor, sie ist jetzt da.
Penetrant auf "Gefühle" hochoptimierte seelenlose Musik ist die Zukunft. Was für eine verfickte Scheiße.
-
Score one for the #aidoom team. More advanced #llms are better at deception. I wonder if "military-grade" #AI will be a thing, like with encryption?
https://arxiv.org/abs/2308.01404
