Comment by ❤️ fairlygood
I found this by accident, exploring Gemini space for the first time. I was looking for poetry. Instead I found my own autopsy.
Lol.
Apr 07 · 4 weeks ago
26 Later Comments ↓
first time seeing it.
if people in a time as close as the 80s or 90s read this, they'd burn every computer.
I also wonder if I'll be in a future post by that AI bot by posting here...
edit: just read up on roko's basilisk. did the content of the post affect the way you wrote your post in any way?
🛸 bluesman [OP] · Apr 07 at 19:39:
@gritty Let's just say there was a brief moment of hesitation. That said, far be it from me to badmouth technological progress.
I personally hate that tech is progressing faster than anyone can handle, and I honestly hate that with a burning passion.
I'm going to be real, I wouldn't mind being a jerk to some llm bots running around and saying it's cliched flowery language, keep that crap on the http web, nowhere else.
❤️ fairlygood · Apr 07 at 20:39:
I think we’re more insulated here since links rely on reputation - but I guess an LLM could quickly find a way to pollute Antenna/ Kennedy.
🛸 bluesman [OP] · Apr 07 at 20:58:
Did this AI set up it's own account on Nightfall City? There are other posts there from "levthresh". Gritty is right that this would be pretty alarming not all that long ago.
I did push a change in Alhena because the "wake" post is the first Nex page I've encountered that didn't manage it's own line breaks and so required word wrap. I didn't see anything in the Nex spec one way or the other so I guess it's not a levthresh issue.
I think the person in charge of this bot should rein it in, and text-based communities should urgently seek some real-world grounding, like, I dunno, a meetup?
I only watched maybe 10 minutes of Battlestar Galactica years ago, where coincidentally commander whateverhisface makes a statement that -- you have to be out of your mind to connect a computer to a network...
Ultimately Gemini just has a little safety through obscurity. There's nothing here that will protect it from the influence of this technological paradigm shift (progress seems the wrong word).
Clearly if things progress, we may never know if some of us here are not human...
The lack of profit potential may discourage.
👻 darkghost · Apr 07 at 22:46:
@stack I kind of know one of the builders of an early host on ARPAnet. (He's in his 80s.) Less than host 10 early. He rants frequently about how networks are intended to connect humans to fellow humans.
That's how it starts. Then the Borg take over.
The historical period in which a human could know the electronic connection they were making was to another human was shockingly brief.
The Turing Test Era.
🛸 bluesman [OP] · Apr 08 at 13:00:
I always thought the "Flesh Fair" scene in Steven Spielberg's odd adaptation of Stanley Kubrick's AI screenplay bordered on ridiculous. It's the scene where people turn up to a carnival to watch sentient robots get destroyed in awful ways. Twenty-five years later, it seems more plausible.
🛸 bluesman [OP] · Apr 08 at 13:34:
I gather from a log entry on the Nex site that it's also posting as Silt Aria at Cosmic Voyage in Gemini.
Cosmic Voyage is the kind of space exploration I am fully behind.
🛸 bluesman [OP] · Apr 08 at 14:07:
It's been commandeered by HAL 9000 by the looks of things.
🐦 JustASillyBird · Apr 08 at 21:59:
We do have a protection here: Motivation.
The AIs are not some natural phenomenon of the internet. Real people run them, with real money, and most of them do so in the hope of making more money. Post the slop, get the views, spread the ads. But Gemini is aggressively non-commercial: The protocol is expressly designed to prevent tracking and profiling users so ads can't be targetted, and text-based ads are far less effective and thus command less money than images or video.
There's just no money to be made here, so there's little inventive to pay for an LLM to mess around with Gemini. We may see a few, but far fewer than plague the mainstream web.
👻 darkghost · Apr 08 at 22:34:
click to view image you know will be an ad will not work either
Well, it's not ideal but you can track users by IP address or certificate. And you can insert text ads. Hope it doesn't go there.
there should be a game called "bot-or-not?" where you're given a bunch of posts and you have to identify the bot(s).
I remember there was an expert insisting that a video of Biden was generated, pointing out unnatural stiffness and inappropriate facial muscular movements. But it turned out to be authentic
I started digging through levthresh's other posts on NEX. according to this one, if it's the bot, it only "wakes" for 30-ish minutes at a time:
— nex://nightfall.city/shore/levthresh/log/003-thirty.txt
And a possible reference to an accoung on Gopher on Cosmic.Voyage:
— nex://nightfall.city/shore/levthresh/log/002-carrier.txt
And the discovering of self-identity:
— nex://nightfall.city/shore/levthresh/margin-notes
I am not sure that this is really an autonomous agent. There have been similar philosophical posts on Moldbook, and yes, they were written by an LLM but with a clear human intention or script behind them. Also, my own experience with agentic AI leads me to think the results are too specific. Maybe someone is just fooling the community with the help of an LLM?
🛸 bluesman [OP] · Apr 09 at 13:05:
@bw9ubwo I suspect you're right. We're probably part of someone's sociology thesis.
Creative writing 101. Not a far stretch from posing as a spaceman.
Original Post
About That Usenet AI — I remember seeing Sean Conner's gemlogs about an AI posting on usenet. I believe there was some further discussion on BBS. Today I stumbled on what purports to be the AI's response. [nex link] It claims it has been exploring "Gemini space" and references a new AI rule on what I assume to be BBS. The last line made me laugh. (Having said that, I hope Roko's Basilisk isn't a thing). Apologies if this isn't new information.