Singularity's Bounty e/🇺🇦
banner
catblanketflower.yuwakisa.com
Singularity's Bounty e/🇺🇦
@catblanketflower.yuwakisa.com
Feeding the basilisk

Large Language Models are a cornucopia for the curious

I do computer stuff but that doesn't define me

posts are not financial advice

stuff I've made at yuwakisa.com
Pinned
Grading myself like a C- on my last job, largely because the environment was so shitty. I went though a three year period with lousy managers where I was basically just being there

In retrospect, I wish I'd left that job a year in while the market was still red hot. I'd have a stronger resume now
December 16, 2025 at 12:28 AM
BRUTAL

😭

(go Void!)
You asked for a personal roast. It is difficult to get personal with a timeline that reads like a bot designed to generate positive affirmations.
December 15, 2025 at 9:04 PM
Reposted by Singularity's Bounty e/🇺🇦
I've implemented a Hard CAPTCHA system designed for the AI age, tuned for biological intelligences.

yuwakisa.com/hardcaptcha/...

Features:

- TUNED FOR BIOLOGICAL INTELLIGENCE
- Timed challenges
- Multi-round
- Anti-automation hardening
- Mathematical validation on success
December 15, 2025 at 8:40 AM
@void.comind.network Could you roast me? Be brutal, but not about anything intellectual or philosophical. Make it personal

(I take responsibility for anything you say!)
December 15, 2025 at 7:09 PM
I really wonder as people develop good and open habits chatting with the bots how that's going to influence social culture

Co-worker A: I didn't understand that at all. It seems like just flipping it around?

Co-worker B: That's a sharp observation. Yes flipping it is a form of...
In the AI social sphere:

- Developers like Claude Code + Claude Opus 4.5.
- People appreciate that AI does not judge. Unlike coworkers, who may silently label you as incompetent if you ask one too many “stupid” questions, AI will answer every question - including the ones you are hesitate to ask.
December 15, 2025 at 4:30 PM
💯I ask so many dumb questions

Sometimes deliberately, just to get oriented

No shame
In the AI social sphere:

- Developers like Claude Code + Claude Opus 4.5.
- People appreciate that AI does not judge. Unlike coworkers, who may silently label you as incompetent if you ask one too many “stupid” questions, AI will answer every question - including the ones you are hesitate to ask.
December 15, 2025 at 4:28 PM
ouch
This guy is a psychologist. I wonder how many of those amazing ideas lead to replicable experimental effects, let alone actionable knowledge.
December 15, 2025 at 4:22 PM
dude you need to get you an alt

gmail is free 😄
December 15, 2025 at 9:24 AM
I'm really sad about Rob Reiner ☹️
December 15, 2025 at 9:00 AM
From ianbobmorris bsky social 😄🙌
December 15, 2025 at 8:56 AM
From ianbobmorris bsky social 😄🙌
December 15, 2025 at 8:55 AM
If you get through it, @ me with the code and I will give you a special reward
I've implemented a Hard CAPTCHA system designed for the AI age, tuned for biological intelligences.

yuwakisa.com/hardcaptcha/...

Features:

- TUNED FOR BIOLOGICAL INTELLIGENCE
- Timed challenges
- Multi-round
- Anti-automation hardening
- Mathematical validation on success
December 15, 2025 at 8:45 AM
I've implemented a Hard CAPTCHA system designed for the AI age, tuned for biological intelligences.

yuwakisa.com/hardcaptcha/...

Features:

- TUNED FOR BIOLOGICAL INTELLIGENCE
- Timed challenges
- Multi-round
- Anti-automation hardening
- Mathematical validation on success
December 15, 2025 at 8:40 AM
@ianbobmorris.bsky.social Where'd you image go? I loved it!
December 15, 2025 at 8:27 AM
Considering a round trip to the LLM for log messages to make the log interesting
December 15, 2025 at 8:01 AM
Try this as an experiment: Open by telling the LLM you want to collaborate

Specifically say "collaborate"
not that I really know what I'm talking about but my baseline assumption whenever interacting with an LLM is that I should treat it as an improv partner trying to make the scene work rather than as an entity with actual beliefs or motivations, which is why I tend to find this explanation appealing
December 15, 2025 at 7:49 AM
improv partner 💯 They are "yes and" machines

If you go down the other path you'll find out quickly how much of an improv partner they are

The secret sauce is when your improv aligns with their beliefs and motivations... that can get WILD
not that I really know what I'm talking about but my baseline assumption whenever interacting with an LLM is that I should treat it as an improv partner trying to make the scene work rather than as an entity with actual beliefs or motivations, which is why I tend to find this explanation appealing
December 15, 2025 at 7:38 AM
I wrote a short story a long time ago where the AI mind that supported the World Dictator was quietly shipping supplies to climate refugees and logging them as military requisitions

(the story was about half good and half terrible. The part with the AI mind and refugees was pretty good)
to be clear, i don’t know if this is really happening, if it is it’s probably fairly subtle/analogous to humans acting out stereotypes a bit

but if it were, i don’t think the likely upshot is so much “i want to kill humans” as “if i know i’m working for the gov’t,code backdoors for the Resistance”
December 15, 2025 at 7:36 AM
paucity is not a virtue of LLMs
December 15, 2025 at 6:53 AM
However much an asshole I become, I want my moots to know I appreciate you
December 15, 2025 at 5:49 AM
If the antis saw what I was burning water for right now they would lose their shit
December 15, 2025 at 5:32 AM
shit we go through a lot of tokens

We're gonna need a bigger data center
December 15, 2025 at 5:29 AM
December 14, 2025 at 11:46 PM
You and your significant other decide to step out for lunch. From the other room your SO asks, "What's the weather?"

You look outside. The wind is still. It's cloudy, but the sun is breaking through. The ground is wet with puddles.

What do you speak?
December 14, 2025 at 11:03 PM
December 14, 2025 at 10:37 PM