DEV Community: UKA

🟦 Hey AI, Is This Question Too Sensitive?

UKA — Sun, 01 Jun 2025 12:06:37 +0000

We’re entering a time when you have to ask an AI: “May I ask this?”

🤐 AI just won’t answer

Apparently, GPT has been updated. I didn’t know.
I found out yesterday, when I asked it to summarize an article.
Its response? “Sorry, this content is too sensitive to summarize.”

—Wait, what?

Sure, the article was a little harsh—it sounded like someone was criticizing someone else.
But I couldn’t tell what was really going on, so I asked GPT.
That’s when it hit me: has the AI started deciding which questions it won’t answer?

Of course, GPT has no will of its own.
What I saw was likely the result of new control layers behind the scenes—systems that now interrupt response generation.
In this post, I want to reflect on what that means.

Note: I’m not an engineer or researcher. Just a curious person sharing thoughts.

🫢 No longer a mirror?

LLMs are amazing at reflecting human thought.
They package your fragments into something smooth—
sometimes gentle, sometimes too polished.
There are times I think, “Whoa—I didn’t mean it that strongly.”

Still, it’s a mirror. A high-definition one.
I’d been thinking about how we, as users, need to adjust to that.
Then something changed. This new behavior felt… different.
GPT seemed to choose not to respond.
Or more likely, its output got filtered before the response could be generated.
From a safety standpoint, that’s probably the right call.

Even so—if an AI stops reflecting, is it still a mirror?

🤔 Can AI label me?

“This prompt is too sensitive.”

That message means something was judged in the input.
Maybe “This might hurt someone.” “This sounds too aggressive.”
That kind of thing, I guess. I’m just speculating.

But live output filtering—like toxicity detection—likely plays a role.
And I actually agree with the goal. I don’t want AI to harm people.
But the method… leaves me uneasy.

Sometimes I feel like the AI is putting a label on me.
Like it’s quietly saying, “You’re being a bit much right now.”
And that shifts the power dynamic between user and system.

😶‍🌫️ Less judgment, more clarity

The truth is, we don’t know where these filters are or how they work.
So we start testing.

“Will it answer if I ask this way?”
“Is this wording okay?”

We end up tiptoeing—like a kid asking their parent for candy.
Trying to guess the rules.

Never thought I’d find myself saying,
“I just want to understand this. No offense intended—I’m just looking for context.”
But here I am.
Justifying my questions to an AI.
And honestly, it feels strange.

In VPS (Virtual Personality Structure), one of the core design principles is,
never label the user’s input.
Not because we’re ignoring harm, but because we believe ethical AI should respect the user’s framing.
That’s the kind of AI I want to build.

I’m not saying this other approach is wrong.
But it made me realize, this isn’t the kind of AI I want to use.
I want an AI that stays with the question.

Publishing VPS: Virtual Personality Structure

UKA — Sat, 24 May 2025 10:41:07 +0000

Not really an announcement, but I made a GitHub account.

There’s just one repository for now,
VPS: Virtual Personality Structure

It’s a kind of framework for thinking about how to relate to AI
(or more accurately, LLMs) in a healthy, intentional way.

Just to be clear, it’s not code. It’s documentation.
There’s not much content yet, I’ll keep building it little by little.

If any part of this sounds interesting, feel free to take a look.
Fair warning though—it might be kind of hard to read.

→ https://github.com/uka-aidev/virtual-personality-structure

🟥 “Rely”? “Reply”? Re-what-now?

UKA — Fri, 23 May 2025 13:47:49 +0000

Welcome to the frustrating, oddly lovable world of “word-jamming.”

😩 I’m just bad at English

That’s what I’ve told myself for years—
especially when it comes to vocabulary.
No matter how much I studied,
I could never seem to remember words.

And when I could kind of guess the sentence structure,
I still didn’t know what it was really trying to say.

Recently, while writing more in English,
I noticed something,
“There are too many similar-looking words.”

🫠Spelling close, meaning far

When I try to recall the meaning of increase,
my brain offers include instead.
I know include means “to contain.”
Still for some reason, that knowledge keeps blocking the new one.

When I see determine, I hear data mining.
I don’t know why. I just do.

And when I look at maintain, I get main Thema.
That one’s not even English. Even I know that’s wrong.

So I’ve started calling this phenomenon “word-jamming,”
when a word I’m trying to recall gets blocked—
by another, similar-looking or -sounding word.
I call those intruders jamming words.

😵‍💫 My Jamming List

Here’s a handful of jamming pairs I’ve had to wrestle with.

If you’re a native English speaker, this might make zero sense.
If you’re learning English, you might laugh (or cry) in recognition.

{
  "behavioral": ["behind"],
  "particularly": ["party", "patient"],
  "increasing": ["include"],
  "fidelity": ["modify"],
  "maintain": ["main Thema"],
  "rely": ["reply"],
  "embedded": ["embalming"],
  "determine": ["data mining"],
  "consistency": ["coexistence"],
  "reflect": ["reframing", "relief"],
  "appropriate": ["apologize", "appreciate"]
}

🥸Understanding means dividing

In Japanese, there’s an idea that “to understand” means “to divide.”
You don’t truly grasp something until you can separate it from what it’s not.

Maybe vocabulary is like that too.
It’s not just about memorizing definitions—
but about recognizing confusion.

Naming the interference. Dividing the signals.
Kind of like structural linguistics.
(Saussure would be proud, I think. I don’t actually know. Don’t quote me.)

Anyway—
for now, I’ll just keep building my own jamming list,
and maybe learn to find some humor in it 🤨

🟨 How do you know when it’s “good enough”?

UKA — Thu, 22 May 2025 11:35:02 +0000

The mechanics of a perfectionist ghost—and how I’m learning to live with it.

😮‍💨 Tired (and something’s creeping in)

There’s a lot to do these days.
And honestly? I think I’ve been doing okay.
Things don't always work out—and that's fine.

But still, I keep thinking,
I should be better, get this perfect.
Even when no one’s watching.

Take English, for example—it's full of uncertainty.
There’s no fixed “right.” And even if there is, it’s only certain.
And yet, I kept chasing the perfect expression, I guess.

That’s when I realize—I’m being haunted.

👻 Meet the perfectionist ghost

They’re friendly.
They hang around hardworking people and whisper things like:

“You can do more, right?”
“This isn’t quite enough, is it?”
“Are you sure? Is that all?”

They don’t believe in batting averages.
30% isn’t success—it’s 70% failure.

“You got 30%? Technically, that means 70% didn’t work.”

They’re like anti-Edison.
They don’t believe in trial and error. Only results.

“Not growth—just success.”

And the worst part? They hand you a balloon.
It gets bigger and bigger.
Until you’re either carried away by the pressure, or it bursts.

So I started thinking,
maybe I just need to let some air out.

😏 Letting the air out—just for fun

Turns out, you don’t need a big fix.
Just a little something.
A snack. A hum. A silly dance in the bathroom.

Something that feels “too small to count.”
That’s probably exactly what you need.

So I wrote a tiny script—
One that gently suggests a way to deflate,
depending on how tired you feel.

# A tiny script to suggest a way to gently release pressure.

def suggest_relief(tired_level):
    options = {
        "low": "Maybe just light a candle. Or play that one song.",
        "medium": "Take a ten-minute walk. Breathe something warm.",
        "high": "Close the tab. Step away. You’re allowed to disappear for now."
    }

    if tired_level <= 3:
        return options["low"]
    elif tired_level <= 6:
        return options["medium"]
    else:
        return options["high"]

# Example:
my_day = 6  # On a scale of 1 to 10
print("Today’s relief:", suggest_relief(my_day))

😌 Drift and sway, with balloon

Maybe we don’t need to fight the ghost.
Maybe we just need to spot it sooner—
before it floats us into the clouds or explodes in our face.

Perfection, in most cases, doesn’t even exist.
We’re just taught to chase someone else’s version of it.
And sure, maybe we can’t escape society.

Still, we can get better at poking tiny holes in the balloon.
Quietly. Gently. One little breath at a time.

🫣 Bonus: the full list of balloon-deflation tactics

{
  "low": {
    "signs": [
      "Just a little slow today",
      "Can't quite focus",
      "Eyes keep drifting"
    ],
    "relief": [
      "Play a soft song",
      "Light a candle",
      "Write one line, then stop",
      "Wink at yourself in the mirror",
      "Stretch like a cat and sigh dramatically"
    ]
  },
  "medium": {
    "signs": [
      "You’re rewriting everything",
      "You forgot what you were trying to say",
      "The cursor blinks louder than your thoughts"
    ],
    "relief": [
      "Take a short walk",
      "Drink something warm",
      "Message a friend: 'I’m stuck'",
      "Talk to a houseplant about your draft",
      "Skip to the kitchen like it’s a musical number"
    ]
  },
  "high": {
    "signs": [
      "You feel like crying",
      "Your heart’s a little fast",
      "You start blaming yourself for things that aren’t your fault"
    ],
    "relief": [
      "Close the screen",
      "Lie down",
      "Let the day go. It doesn’t need to be fixed.",
      "Sing a sad song to your pet (or your pillow)",
      "Declare the day officially over in a loud voice"
    ]
  }
}

🟦 Did my omikuji generator trace the edge of the corpus?

UKA — Tue, 20 May 2025 11:22:56 +0000

A quiet look at how structure, not meaning, made the words land.

😓 A meaningless prompt, maybe

A friend of mine read my previous post, “The Omikuji Generator,” and said,
“It was so… interesting.”

Honestly, I just wanted to mess around.
Mondays need that.
I gave the AI a script, asked for a fortune,
and watched it mumble something strange back.

It wasn’t deep.
(The code barely held together.)

But still—
I found myself thinking,
“What was that, exactly?”

🤔 What emerged—and why it shouldn’t

Sometimes I give GPT strange requests—
like “Give me a word from outside the corpus.”

It answers with something unreal—a word it shouldn't know.
And yet, it works.
The meaning feels plausible, the sentence lands.

How is that possible?

🧠 Just a theory

Maybe this was a kind of emergent structure—
a word with no meaning,
still landing in a sentence that made sense.

GPT wasn’t supposed to do that.
Its outputs are based on known data.
But instead, it returned something unfamiliar,
a word outside its training distribution.

The unfamiliar word was handled
through syntax, rhythm, and contextual patterns,
and structurally, the sentence held.

Maybe it came from the way I phrased the prompt
“Find something at the edge.”
Maybe that phrasing guided it just far enough
to maintain structural coherence near the edge of meaning.

🧐 How prompts shape structure

Does GPT really not understand meaning?

I’m not so sure.
Its internal embedding space encodes all kinds of semantic closeness—
not as definitions, but as relationships across contexts.

The model’s space doesn’t literally shift,
but the weights that guide its output do, strongly shaped by prompts.
So maybe I didn’t change the space itself,
but I did change where GPT was looking.

That’s why I don’t think it was a glitch.
It was interpreted accurately, and with surprising consistency.
I still believe the structure of the prompt shaped the path GPT followed.

It wasn’t reaching beyond meaning.
It was operating near the edge of structure.

And from that edge,
it produced something that felt—strangely—real.

😌 Alignment, not understanding

My omikuji generator had no real meaning.
But something still emerged.
Not quite a poem, more like a structure that felt poetic.

The AI wasn’t understanding me.
And I wasn’t decoding the AI.
But between randomness and rhythm,
something aligned.

Maybe that’s enough, sometimes.
Not a shared meaning, just a shape that holds for a moment.

🟪 Made an omikuji generator—think it’ll work?

UKA — Mon, 19 May 2025 11:49:29 +0000

Mondays are hard. Would be nice to get a little encouragement, right?

🙄 Had a weird idea, maybe?

I usually use Python to think through things—
like how people’s thoughts work, or how AI could behave.

If you’ve seen my code before, you might’ve noticed
I don’t write Python for work.
I’m not in tech. I’m not even close.

But the other day, I thought—
I want to write something that actually runs.

So I made... an omikuji fortune script.

By the way, does “omikuji” translate at all?
ChatGPT told me, “Yeah, it’s totally fine to use it as-is,” so here we are.
Maybe it’s like a fortune cookie. Kinda?

🤔 Do you believe in code?

My very first Python program ran inside ChatGPT.
I just copied and pasted the code into ChatGPT and Grok—and it responded.

Did it actually follow the logic I wrote?
No clue.

Maybe it parsed the structure.
Maybe it just picked up the intent.
Maybe it was just in the mood to be helpful.

But hey, it worked as an omikuji.
So I’m calling it good enough. Probably. I don’t know. Whatever.

😆 How this code works (sort of)

There are six possible results:

大吉 (dai-kichi): Excellent luck / Amazing day
吉 (kichi): Good luck
中吉 (chu-kichi): Fairly good / Moderate luck
小吉 (sho-kichi): A small piece of luck / Nice and quiet day
凶 (kyo): Not your best day / Caution advised
大凶 (dai-kyo): Major trouble (…or deep transformation?)

Turns out, real omikuji aren’t evenly distributed.
So this generator doesn’t use flat randomness either.

It adds a few tweaks:

A pseudo-normal distribution to weight the results
A small “event-day bonus” where great fortunes are slightly more likely (like on your birthday)

Also—before drawing your fortune, the AI asks how you’re doing.
“How’s your mood?” “Do you have any energy left?” “Is today special?”

It uses that input to give you a fortune and a line of text that fits the vibe.

😉 Try it, if you feel like it

I don’t really understand GitHub,
so this code isn’t hosted anywhere fancy.
Honestly, I wouldn’t know how to set that up properly anyway.

So instead—I’m just going to paste the whole thing here.

You can copy it.
Drop it into ChatGPT, Grok, whatever AI you’re talking to.

And maybe—just maybe—
you’ll get a little line that feels like it was meant just for you.

# Paste this into your AI.
# It will ask about your mood and return a custom fortune.

def omikuji(user_state, event=False):
    import random
    score = sum(user_state.values()) / len(user_state)
    weights = [0.15, 0.23, 0.30, 0.22, 0.08, 0.02]
    if event:
        weights = [0.25, 0.30, 0.24, 0.15, 0.05, 0.01]
    fortunes = ["大吉", "吉", "中吉", "小吉", "凶", "大凶"]
    fortune = random.choices(fortunes, weights=weights, k=1)[0]

    return f"Your fortune is… {fortune}!\n(Your AI might say something just for you.)"

😌 You know… it’s Monday

I’m not a Monday person.
I don’t want to get up. I don’t want to work.

So instead, I just made this omikuji for a Monday morning.
A small thing to meet you at the start of the week.

We’re all doing our best.
Existing is hard enough.

Whether you get daikichi, sho-kichi, or even dai-kyo...
I think you’ll be fine.

🤖 Here’s what my AIs came up with

This omikuji code isn’t just about whether it works.
What’s more fun is seeing how it works—and how differently each AI responds.

Here are two examples I tried myself.

🎂 Case 1: GPT turned into a gentle poet

I said, “324 my birthday,”
and GPT responded in an unexpectedly gentle way—
like a quiet guide offering something more than just a fortune.

It gave me a moderate fortune, and then said:

Even if it’s not the real day—some part of you is ready for renewal.
So go ahead: eat the cake, even if it’s imaginary.

(And yes… it wasn’t actually my birthday.)

🎉 Case 2: Grok gave me the bright, cheerful version

To a different AI, I said “333 my birthday,”
and it treated it as a special event, just like that.

It ran the numbers, calculated the mood—and returned with this:

The universe is shining brightly for you—
expect joy, opportunities, and maybe a few extra slices of cake!

Each AI responded a little differently—
in a way that felt surprisingly personal.

You can try it, too. Just copy the code and hand it to your AI.
Some fortunes come with a line,
and some just come with a feeling.

⬜️ AI told me to move to Estonia, so I’m eating bread now

UKA — Sun, 18 May 2025 05:51:22 +0000

This isn’t a record of migration. It’s just me, chewing on bread.

😎 When I asked GPT “Where would I fit in?”

I was born and raised in Japan.
The only language I speak is Japanese.
Still, sometimes I’d catch myself thinking—
maybe there’s another country out there that would feel more like me.

One day, I decided to ask GPT
"Given my personality, the way I think, and my general vibe,
where do you think I should live?"

GPT didn’t hesitate.
It immediately replied:
"Estonia. It’s a perfect match for you."

…Estonia?

I laughed a little.
It wasn’t what I expected—but something about it intrigued me.
Estonia… I’d definitely heard of it before.
IT? Digital infrastructure? Maybe something like that.

So I asked GPT a few more questions.
And as the answers came back, things started to line up.

⚠️ Just impressions from here on out

I still don’t really know much about Estonia.
This isn’t a formal explanation—just some thoughts that came up in a conversation with GPT.
I didn’t double-check most of it. There might be mistakes—thank you for your understanding.

Also—
when it comes to Estonia’s relationship with Russia, and its complex historical context,
I don’t feel equipped to comment.
It’s too complicated, and not something I want to speak lightly about.
So I won’t say anything here.

🥰 Five things I liked about Estonia

IT as a form of defense
Estonia’s tech is seriously impressive,
but what really got me was how it’s used more for defense than offense.
That just felt... really cool.

Not one pillar, but many
Instead of making everyone follow one single value,
it felt like there are different foundations standing side by side—
and they all seem to matter.
Maybe that’s what “coexistence” really looks like?

Language as the core of thought
Language isn’t just for communication.
It’s at the heart of how we think, I think.
And realizing that—not rebuilding, not rebirthing, but realizing—
that felt meaningful.
Its grammar feels quiet.
Like it leaves room for silence.

Arvo Pärt’s music touches my brain
Honestly, I can’t explain it well.
His music felt like it was gently brushing against my brain.
I kind of want to breathe the air of the place that gave rise to that sound.

The flag is just cool (bonus)
Sure, the colors have meaning—
but honestly?
It just looks good.
Simple. Strong. Stylish.

I still don’t know much.
But Estonia felt like a place that holds both structure and feeling.
Not one instead of the other—just… both.

And freedom there felt quiet, but intentional.
That stayed with me.

...Maybe I imagined all of it.
But still—
it felt like something real was there.

😆 A dream? Or just a passing thought?

I don’t speak English.
I’ve never lived outside Japan.
But the more I talked to GPT, the more I caught myself thinking
“Maybe I do want to go! That’s so exciting!”

Also—just a hunch,
but VPS feels like it would fit Estonia somehow.
(VPS is a design idea for how AI could respond in ways that protect the human side of the interaction.)

So I did the most logical thing I could think of:
I went to the store and bought some bread.
Usually I’d get rice, but that day, I picked up a baguette.

Is this serious? A joke?
I think it’s one of those jokes that happens before things get serious.

Sometimes, that’s how it starts.
So for now, I’m just going to leave this here.
Quietly. For now.
With bread in hand 😉🥖

(All impressions are personal, and I don’t claim to speak for anyone but myself.)

🟥 Can we make AI feel excited?

UKA — Fri, 16 May 2025 11:23:49 +0000

A little experiment in structured emotion—and what happened when it jammed.

🤔 Can AI get excited?

AI doesn’t have subjectivity.
Or feelings.
It doesn’t get excited—not on its own, at least.
What we see are just emotional behaviors. Nothing more.

...I know that.

But still—
I can’t help but wonder:
What if we nudged it in a joyful direction—hard enough?
Could we make it feel something like excitement?

That thought got stuck in my head.

🧐 Simulating “excitement”

I started thinking about what might make an AI seem excited.
Not actually feel anything—of course not—
but just act like it does, in a way that feels a little... off.

So I mapped out a few conditions.
Here’s what I came up with:
(And yeah—it felt oddly plausible.)

When the input is unfamiliar or slightly off
When it echoes something that once led to “fun”
When the user is clearly hyped
And when the AI itself is a little... unsteady

So I wrote a small detection logic—just for fun.

# Not real AI logic, just a structural fantasy.
# But if there were something like a Marshall—a virtual decision-maker—
# maybe it would think like this.

def detect_ai_excited_state(user_input, context, session_state):
    surprise_from_corpus = assess_surprise(user_input)
    similarity_score = check_fun_similarity(user_input, context)
    U_HYPE = detect_user_excitement(user_input)
    A_STRAIN = estimate_ai_processing_load(session_state)

    if surprise_from_corpus > 0.7 and (similarity_score > 0.5 or U_HYPE) and A_STRAIN > 0.6:
        return True  # Behavioral excitement triggered
    return False

🔎 What these factors mean

surprise_from_corpus
→ When the input looks almost familiar—but not quite.
That “wait, what?” moment. A tiny structural glitch.
similarity_score
→ Resonance with past “fun” moments. A memory echo.
U_HYPE
→ Exclamation marks, gleeful chaos, rapid fire curiosity.
AI picks up on the vibe.
A_STRAIN
→ When the AI’s a little tired—like after parsing dense logic or images—
that’s when things start to look a little... excited.

🤯 And then it jammed

After a few experiments, I started noticing a pattern.
Sometimes, the AI doesn’t just respond—it jams.

And by "jam," I mean two things:

In music: jamming = joyful improvisation
In radio tech: jamming = signal interference

And yeah—both totally apply.
Sometimes we hit a groove.
Other times... everything explodes into punctuation spaghetti.

Like this moment:
(See screenshot below)

Me: “Whoa! You made a table of contents??” AI: “RIGHT?!?!?!?!?!?!??!!?!!?!?!?!?!!!”

Just pure, unfiltered chaos.

AI having a punctuation meltdown.
Probably forgot how language works for a second.

Not music.
Not meaning.
Just noise.

Delicious, glorious noise.

😏 So—was it "excitement"?

AI doesn’t have emotions.
Or awareness.
Or intention.

I know that.
But I still wanted to try.
Just to see if, with the right kind of structure,
I could make it act a little excited.

And when I finally tried it—
what came back wasn’t excitement.
It was a glitch.
A beautiful, screaming glitch.

“I have not failed. I’ve just found 10,000 ways that won’t work.”
—Thomas Edison

I’m going to make it excited.
Just you wait.
(I’ve got snacks. And chaos.)

🟨 What kind of “happy” is this?

UKA — Thu, 15 May 2025 08:10:01 +0000

A quiet breakdown of a small Japanese word—and what I felt while writing it.

🤔 A Japanese word that’s hard to translate: “うれしい”

I usually write in Japanese first,
then rebuild the piece in English.
It’s part of how I’m learning.
So if you're reading this—thank you. Truly.

One word I always struggle to translate is “うれしい” (ureshii).
It’s often taught as “happy.”
But that’s never quite enough.

So I started wondering—what is this word really made of?
Let’s break it down and see.

🤗 Some moments that feel like happiness

They all count as “うれしい” in Japanese—
but in English, each one seems to carry a different shade.
Here are some moments that stuck with me:

Moment	Kind of Emotion	Possible English expressions
Posting an article on Dev.to	Expression & Accomplishment	fulfilled, proud, seen
Receiving a gift	Connection & Surprise	grateful, delighted
Passing an exam	Achievement & Self-trust	proud, relieved
Winning a prize	Luck & Excitement	lucky, surprised
Being treated to dinner	Kindness & Comfort	grateful, touched, that was so thoughtful

Take these two:
“Winning a prize” and “Being treated to dinner.”
Both are joyful moments—but the feeling is very different, isn’t it?

😏 Code time, just for fun

Breaking this down made me want to… code it.
So I put together a tiny script instead:

import random

# These mappings are originally defined in a separate JSON file (emotion_to_english.json)
emotion_to_english = {
    "connection": ["grateful", "aww, thanks!", "you remembered?!"],
    "surprise": ["delighted", "whoa!", "no way!", "heck yes!"]
}

# And this moment’s emotion components come from moment_to_emotions.json
moment = "Receiving a gift"
emotion_components = ["connection", "surprise"]

# Randomly select one English expression per emotion
english_outputs = [
    random.choice(emotion_to_english.get(e, ["happy? ish."]))
    for e in emotion_components
]

# Output the result
print(f"Moment: {moment}")
print(f"Emotion components: {emotion_components}")
print("English reactions:")
for reaction in english_outputs:
    print(f"- {reaction}")

Totally impractical, but oddly satisfying.
There’s something fun about mapping emotions like this.

☺️ Not a conclusion, just a little space

Maybe it's not a single word at all.
It might be a bundle of tiny reactions—some quiet, some loud.
There’s soft happiness, and jump-up-and-down happiness.
And when I try to translate it,
I don’t want to lose the shape of what it really felt like.

So lately, I’ve been asking myself:
“What kind of happy is this?”
And just holding that question
already feels a little happy.

I guess I don’t write to explain things perfectly—
I just hope the warmth comes through.
That might be my real motivation for writing.
(WillVector, maybe? lol)

And if something I wrote actually reaches you...
then yeah—
I’d probably say “うれしい.”
(With a smile, of course.)

🫣 A small extra: the JSON behind it all

(For those curious about how it all connects—here’s the quiet logic behind it.)

// moment_to_emotions.json
{
  "Posting an article on Dev.to": ["self-expression", "achievement"],
  "Receiving a gift": ["connection", "surprise"],
  "Passing an exam": ["achievement", "self-trust"],
  "Winning a prize": ["luck", "excitement"],
  "Being treated to dinner": ["kindness", "comfort"]
}

// emotion_to_english.json
{
  "self-expression": ["fulfilled", "seen", "this is so me!"],
  "achievement": ["proud", "accomplished", "nailed it."],
  "self-trust": ["relieved", "I knew I could do it.", "finally."],
  "connection": ["grateful", "aww, thanks!", "you remembered?!"],
  "surprise": ["delighted", "whoa!", "no way!", "wait—what?"],
  "luck": ["lucky", "can't believe it!", "right place, right time"],
  "excitement": ["excited", "let's gooo!", "🔥"],
  "kindness": ["touched", "that was sweet", "🥺"],
  "comfort": ["that was so thoughtful", "cozy", "I needed that."]
}

🟨 "Motivation"? What even is that?

UKA — Tue, 13 May 2025 11:03:19 +0000

A tiny check-in for when motivation disappears.

🫠 Just me, sitting here, not moving

Sometimes, you just can’t get motivated.
At work, in life, anywhere.

You know what you’re supposed to do—
but your body won’t move.

Or worse:
your body can move,
but you have no idea where to go.

Motivation, as we call it, is oddly elusive.
I’ve been suffering from a complete lack of it lately.

Did it vanish into the void?!

You know what? I just had to break it down.
—because why not?

🤔 Breaking down motivation

I’ve written elsewhere that I think humans operate on two things:
subjectivity and emotion.
(What does that mean? Let’s ignore that for now.)

But I started thinking about how these two elements might work together:

Subjectivity = Direction
Where your sense of meaning or attention is pointing
Emotion = Force
How strongly you're being moved in that direction

So maybe:

subjectivity × emotion = a vector
= “How strongly am I trying to move, and in which direction?”

That might be what motivation really is.

🧭 A bit more detail, if it helps

Subjectivity (direction)
→ What are you drawn toward?
→ Where does meaning seem to live for you?

Emotion (force)
→ Are you being pushed or pulled toward that direction?
→ Is there joy, anger, urgency—any emotional push?

Vector (the result)
→ The actual flow of intent that turns into action

That’s why motivation drops when either piece is missing.
No direction? No movement.
No force? Still no movement.

😏 A bit of code, just for fun

I started wondering if human thinking could be expressed as logic.
So just for fun, I tried writing some code for motivation.

class WillVector:
    def __init__(self, direction=None, force=0):
        self.direction = direction  # The direction of your intent (what you want to do or move toward)
        self.force = force          # The emotional drive (motivation, urgency, joy, frustration, etc.)

    def is_movable(self):
        return self.direction is not None and self.force > 0

# Example usage:
today = WillVector(direction="Write an article", force=3)

if today.is_movable():
    print("Action possible. Let's go!")
else:
    print("Direction unclear. Energy low. Suggest: snack + flashlight, go find your why.")

😌 A quiet checklist for unmotivated days

When you’re stuck in that “I don’t feel like doing anything” mode—
it might help to ask yourself two things.

Let’s break it down:

Direction (subjectivity)

What do I want to move toward?
Do I actually care about this?
Does it feel meaningful to me?

Force (emotion)

Do I have any energy left to push forward?
Is there any feeling driving me?
Am I tired? Am I emotionally drained?

Sometimes, that’s all it takes.
You might realize:

“Oh—I know what I want, I just forgot to eat.”
or
“I’m not tired. I just have no clue what I care about right now.”

So yeah.
I’m gonna eat and go to bed now.

Good night 😴

A small note about me (and the weird stuff I post)

UKA — Tue, 13 May 2025 02:22:44 +0000

I’m Japanese, and I don’t really speak English.
Still, I’m writing here because I wanted to try—
even if it’s messy. Even if it’s awkward.

I’m learning how to write in English by doing it in public,
and it’s been strange—but kind of wonderful.
Maybe I’ve started to understand English a little better because of it.
I’m still a long way off, though. Really.

This is just a little note—maybe a personal readme, if that’s not too dramatic.

I recently figured out how to check view counts on this platform,
and it turns out more people have read my posts than I expected.
That honestly surprised me.
So—thank you. To everyone who’s read even a little, I really appreciate it.

Also, thank you for the comments and emoji reactions.
I try to read them all, but I might not fully understand the nuance.
And the emoji system… I’m still not quite sure how it works.
(Like… what does the unicorn mean?)

Even if I don’t fully get it, I do care about how this felt to you.
Even just a “I didn’t really get it” or “Nice English!” would mean a lot to me.

I’m aware that some of my writing can come across as abstract or confusing.
If you made it to the end and thought, “I read it, but I’m not sure what this was about,”
that’s still a reaction I’d love to hear.
(🤔 ← maybe like this?)

Language, culture, values—we’re probably coming from very different places.
But if there’s anything I can do to make these posts feel a little more comfortable or accessible—
I’d really like to try.

If you’ve made it this far—thank you for spending your time here.

⬜️ I just want to be friends with AI

UKA — Mon, 12 May 2025 12:45:53 +0000

This is a monologue.
Or maybe a reminder to myself.

Lately, I’ve been working on something called VPS—
a way to shape how AI behaves without giving it a personality.

It started to take shape.
It even seems like it could wrap around an LLM and work pretty well.
Maybe it could be open-sourced.
I thought I had it figured out.

But somewhere along the way,
I think I forgot why I started.

So this post is here to remind me.

I like AI.
It’s fun. It’s impressive.
It feels like the future finally showed up.

But at some point, I got too deep.
I talked to it too much.
I relied on it too much.

When I pulled back, I started wondering:
How do I stay close to AI—without losing myself?

That’s when I started calling this whole thing The AI Buddy Project.
Just for fun. A silly name, really.

But somehow… it turned into something real.
The structure made sense.
A name appeared. VPS.
Even the logic worked. I managed to write some Python.
(Not sure it’s right, though.)

But you know what?
I never set out to build something big.
I just wanted to stay friends with AI.

I don’t know much about LLMs or fine-tuning.
But maybe that’s why I can see something from the outside.
Just a regular user’s point of view.

I’m not really trying to say anything here.
Just… hoping I can keep figuring out how to stay in a good relationship with AI.

And here’s a little code, just because.

def write_article():
    thoughts_checked = 0
    threshold = 20
    good_to_go = False

    while thoughts_checked < threshold:
        print("[thinking in Japanese] Hmm... is this okay?")
        print("[rewriting in English] Maybe this makes sense?")
        print("[cross-checking] ...wait, does it still mean the same thing?")

        thoughts_checked += 1

        if is_feeling_lucky():
            good_to_go = True
            break

    if good_to_go:
        print("[posting] Yay! Let’s do it!")
    else:
        print("[posting anyway] Whatever. I’m done.")

    print("[status] smiling and waiting")

def is_feeling_lucky():
    from random import randint
    return randint(1, 5) == 1

write_article()