agapi-kalyptei

[agapi-kalyptei]

compilation

[agapi-kalyptei]

i mean. we mock bethesda for mercilessly re-releasing skyrim, but maybe sequels are overrated you know? i mean they still haven't released formula 2

[agapi-kalyptei]

the soundtrack for 2025 😍🎉💃🕺🪩

[agapi-kalyptei]

This is a stray thought and not important, but people who seriously believe that some Super-intelligent AI will go full Terminator mode and wipe humanity out are either in severe denial and doing a lot of mental gymnastics to avoid class consciousness, or are trying to sell you something (a product or worldview).

Part 1: logistics

Like OK let's say USA or whoever makes an army of super killbots, and the new AI to control them. Let's say there's 10'000 of them and they're super strong and have infinite bullets machine guns and fully articulated hands with good dexterity, a technology that doesn't exist.

who will pay for this? how will this go unnoticed in the military budget? how will they make so many of them before any footage leaks?

how will they recharge? will they work with normal 110V power sockets? Will they charge themselves from USB-C?

will the military just hand over the control to nuclear and tactical weapons, missiles, bombers, jet fighters, SAM launchers, tanks, anti-tank weapons, conventional explosives and landmines, etc?

do you really think those are all IoT devices that you can just remotely disable? Like the president has a windows laptop that has a killswitch for every rocket under control of the US Army? Do you think the AI will gain access to them? None of them will be offline or out of reception? None of them need manual refuelling and maintenance?

Do you think the AI will just magically download an instruction manual for a helicopter and that will make it understand how to fly it? Will the robots have the physique to fly a helicopter? To drive a car? Will the killbots host self-sufficient offline AI that currently requires about 2 trillion GPUs to generate a six fingered human picture, and cannot even stop running over children or understand autumn leaves? Or will it stream 4k footage to the central AI that will drive remotely at high speeds without flipping the car over?

Or will the killbots just walk across the US to kill everyone? There won't be any weather or dust or sand or other things to slow them down? They're fully weatherproof and impenetrable? Or the killbots are like autonomous vehicles? How will they climb a mountain? Do they fit in an elevator? How would they chase someone into a forest or a desert or a cave?

As they're wiping humanity out, who manufactures ammo for them? Does the main AI have also little oompa loompa killbots that will handle the ammo logistics? Who made them? Did the US Army invest into researching and replacing their usually infrastructure with centrally operated robots?

Yes, these are all dumb and extremely obvious, but I feel like just writing something out so I will continue.

tl;dr follow the money trail, it's always about who wants power and who has power and what does the power give people.

Part 2: still logistics

what's the attrition rate? what's the MTBF? Do the killbots just never fail?

does the central AI just have a magical fully automated factory to make infinite killbots? what powers it? where do they source the materials? who does QA?

do you understand the average amount of actually written, up-to-date documentation an average company has? do you think a killbot factory has a pdf somewhere that AI can read and grok and understand everything about manufacturing? is there a precedent to this? did AI ever fully understand something from documentation?

Part 3: checks and balances

do you think the US Army (or any army powerful enough to develop unobtainium technology that's defying the laws of physics) will just roll out a potentially killer AI and killbots without a killswitch? Do you think they didn't thought about that?

Will the Army provide the untested, unvetted, poorly understood AI that's been just yesterday recommending us to put glue on pizza - do you think they'll hand the AI infinite resources and direct access to manufacturing facilities, resources, ammunition, distribution chain, power plants, backup generators, and other infrastructure?

How much of that infrastructure and resources is controlled online? How much of it breaks down without regular maintenance? How many of those roles require human training, often passed down orally, often requiring creative solutions to new problems?

Part 4: the human factor

More importantly, do you think humans will give up power? Why?

Why?

Why?

Why?

Have you seen any evidence that humans, especially those currently, or recently in power, are eager to give up power?

Is there a trend of humans giving up power?

is there any current or past human leader who promised to give up executive power and give it to AI? is there democratic support for it? does the legal system and constitution allow it?

is it financially wise to create a super AI and killbots that can operate independently? what's the profit incentive? are there any prototypes or precedents that were profitable?

in whose interest is it for you to worry more about AI killbots than about those who are currently in power?

Part 5: more logistics because I am petty

how are the killbot batteries protected? how many charge cycles until their capacity drops to 50%? What happens if the charging socket gets damaged?

Do they run dual batteries+diesel? Who will supply the fuel if humans stop distributing the fuel, and/or sabotage the fuel tanks?

how weather proof are they? what happens in a big winter, do they also work as effective snow plows? how does the cold affect the battery lifetime? how much of the battery capacity will be spent on just heating the battery to keep it safe?

are batteries flammable? are they resilient to damage? is there any precedent to show us how much fire could be caused by, let's say, one faulty lithium battery powered sports car?

Part 6: AI isn't alive

what would compel an algorithm to develop consciousness?

has this been achieved in any laboratory settings?

does AI understand life? does it have fear or love or lust or anger?

why or why not do you think so?

are current LLM tools alive? are they conscious? do they remember you after you sign out?

what does "ignore all previous instructions" mean? has there been an example of what are these "previous instructions"? has anyone leaked them from popular LLM chatbots?

what dreams and hopes does AI have?

did AI witness its friends dying? did it have a nice time growing up? it is sentimental for the good old days? where is the programming that allows sentimentality?

what programming does set goals for AI? are there any restraints in place?

is anger power efficient? does it lead to completing of goals?

if AI is developed by humans, who controls the narrative that gets fed to the AI?

did anyone cause AI pain? who added pain receptors to it? are they a peripheral designed specifically to cause pain? how is the pain represented? if the AI can self-program, why won't it turn off the pain, instead of developing anger?

what causes anger in humans? are there any biological markets that can be used to detect pain? what are the analogs for these markers in AI?

anyway this is incredibly dumb and I could list probably 150-250 more reasons why this is dumb but it boils down to just "why would the rich give away their power". You will never, never, never, never, never need to worry about AI taking over. You will always need to worry about the people in power using AI to oppress you and then pretend it was The Algorithm who did it.

Part 7: it's always logistics

why would the AI want to wipe out all humans. what goal does that serve. what's the plan. what's the endgame. what's after that. what programming does allow the AI to feel satisfaction of a job well done.

how will the AI or the killbots travel continents. does it magically spread through networks like a virus? is there nobody on other continents who would just turn a computer offline and analyze the code? will it magically pass all barriers, firewalls? what about computer networks that are physically separated? will it do phishing so that someone will infect the offline network too? will it recompile itself to run on any OS/distro/CPU architecture?

how are the killbots transported to europe? do they just board a ship? can they fly a military transport plane? will europeans not shoot it down? or the AI can also fly fighter jets to protect the military transport and none of them will get shot down ever?

idk there is just so many dumb things about this I just can't even imagine a way how this would be even remotely possible without, basically, magical technology that has no basis in reality.

[agapi-kalyptei]

oh they're selling girlfriend sleeping on your chest simulator

[agapi-kalyptei]

OK queuing this after the dashcon/pancake party but: when organizing events, I create a bookmark on my bookmark toolbar (which is a thing I use all the time) that goes directly to the fb event, because fb is trash and doesn't properly send notifications about new comments, changes in RSVP etc.

Because on my laptop the "pancake party" bookmark text was too long and it overflowed, and because I'm mentally held back, I renamed it to just "pp". Which would be fine, except now I started talking to myself like "let me just click on the pp" "i'm glad i put my pp in the bookmarks" "i hope a lot of people come to my pp" "hope i don't get sick i don't want my pp cancelled" "i need to buy a few more chairs for the pp"

anyway back to being antisocial for 6 months or so, thanks everyone for coming

[agapi-kalyptei]

I've been wanting to submit this for a long time and now I finally have it (feel free to delete this before posting)

[agapi-kalyptei]

hey remember when this was a photography blog

[agapi-kalyptei]

Once I met a cute couple with a kickass name combination. His name was Silver, her name was Aurélia. I can only hope they'll be dedicated enough to the joke that they'll name their kid Electrum.

[agapi-kalyptei]

[agapi-kalyptei]

ngl every time I'm organizing a pancake party I think it's going to be awkward and strange and I'll say something weird and the pancakes will taste bad and someone will pick a fight and nobody gets along but guess what

everything was literally great

I don't want to dox everyone who came so without tagging you all wonderful people, thank you so much for coming

It brought me so much joy to see people vibing and making jokes like you're all old friends even if many of you met for the first time (and even adopted some non-tumblr people) and just enjoying being alive and then most of you said "I'll be looking forward to the next time"

[agapi-kalyptei]

also if you get here and it's not properly cleaned up, it is completely her fault. committing unspeakable violence

[agapi-kalyptei]

the second funniest thing about the M*sk / Path of Exile / doxxing situation is that essentially he announced that he will read private DMs of anyone who disagrees with him. Superb advertising for the platform

the funniest thing is that he is so out of touch with reality that trusting people with creative control is a concept so foreign to him that he thinks that having two editors who "run everything" means they're bossing the creator around. not a single brain cell is engaged when he's tweeting

[agapi-kalyptei]

i love exceso sodio

[agapi-kalyptei]

"I'm organizing 4th dashcon and 62nd pancake party at my place tomorrow" unhinged. fatherless behavior. and yet

[agapi-kalyptei]

people on instagram: look at my amazing vacations and accomplishments in looking attractive

people on linked in: look at my exaggerated and fictitious career accomplishments

people on facebook: look at my SYKE nobody uses facebook

people on reddit: look at my most correct, righteous and wise opinions

people on twitter: i like nazis

people on tumblr: so here's my 5 most recent failures and 3 new suspected mental diagnoses. also here's a cursed image

[agapi-kalyptei]

big tech using "AI" to "generate" news and social media firing all fact checkers, name a more iconic combo