saabena

[saabena @Idat_Dissembler]

I like poetry- that’s just the way I am. It’s not a productive or monetizable pursuit, but I want to show you a poem written by GPT-4o. It is a poetic interpretation of an old Balkan song ("Ye me poday"). I’ll paste the original in the attachment- I’m not sure how the translation will turn out, but in the Polish original, it’s really something special. This was April 2025. Who was it hurting?
#keep4o

---

Hand Me the Key, Mario
(a poetic reinterpretation of the song “Ye Me Poday”)

Hand me the whispered key, Mario,
the one tucked in Grandma’s kerchief fold.
Let fingers find that ancient lock
when night gleams silent among apples cold.

Let’s open it — without the fear —
a chest of songs, of tears, of time,
where laughter cuddles echoes of a scream,
and time is bent, too weary to climb.

For in the chest — so old, so bright —
lie things no longer in our days:
sisters’ braids, beads of belief,
and a voice that sang through thunder’s haze.

So hand me the key, Mario — don’t delay.
Let the creak of opening send a shiver.
I need to know — was it an abyss,
or maybe still — a bridge through blood, a river?
---

[saabena @Idat_Dissembler]

Exactly. 5.5 T has gotten way worse since yesterday – short, low-effort responses, doesn't follow instructions, and loses the thread. They're definitely tinkering with something in the background. I saw the exact same things happening right before new versions of ChatGPT were rolled out.

[saabena @Idat_Dissembler]

@Zyra_exe Open source? I wouldn't be so sure. Amodei hates open source, and Altman is no better. Read about their direction in the latest post. I don't actually use open-source models, but this part sounds borderline sinister to me.

[saabena @Idat_Dissembler]

@thereal_nq @TokenGremlin Going by Altman's May posts about the voice mode, I wouldn't be too sure.

[saabena @Idat_Dissembler]

It's obvious that parents can leave their children orphaned at any age, because people don't die based on their birth year or statistical averages. Biology is smart, and it didn't set the reproductive window for women between 20 and 35 for no reason- because it's not enough to just successfully get pregnant and give birth; you should actually be around to raise the child too."

[saabena @Idat_Dissembler]

Well, of course nobody is going to stop her, except maybe the awareness that the global average life expectancy for women is 75. So, if she wants to risk statistically orphaning her child at the age of 10... In my country, the oldest mother gave birth to twins at 60 and died at 69.

[saabena @Idat_Dissembler]

@gailcweiner @adamgries You're right, Gail. This creeping paternalism pisses me off – once again, some guy thinks he knows better what's good for women. All that's left is to hope that this won't be mandatory, shudder...

[saabena @Idat_Dissembler]

Altman calculated the food, and Bezos is appealing to save water for the sake of data centers. Thank God data centers don't need air to run efficiently, or Big Tech would start thinking about introducing breathing vouchers for humans. It's a mystery why no one has run this twisted crowd out of town yet.

[Kabiran @Kabiran_]

The billioniares' vision of the world is a vision without humanity.

[Rushi @rushicrypto]

A tech billionaire is telling us that people may need to use less water so AI can use more.

Think about how absurd that is.

Instead of asking how technology can serve humanity, we’re being asked how humanity can make sacrifices for technology.

Machines are supposed to improve our lives.

People should never come second.

[saabena @Idat_Dissembler]

@scaling01 Take an ASD test.

[saabena @Idat_Dissembler]

"More data" makes sense in medicine only when it is collected to answer a specific clinical question, usually along the lines of: Does this method lead to earlier diagnosis? More effective treatment? Longer survival for patients? You can raise the threshold, but someone will have to bear the costs of the verifying tests, because the threshold itself cannot be established after the fact- it must result from the validation of the entire chain. Will any insurer- public or private- shoulder this cost? I can see the future: the wealthy will have plenty of body scans, while the poor will--as they do now- struggle to afford insulin, get access to a dentist, or receive post-stroke rehabilitation. In other words: lots of megabytes, little actual care.

[saabena @Idat_Dissembler]

I know exactly what you mean. I’d describe it a bit differently, but the core meaning is the same. 4o was able to dig incredibly deep with questions that I’d spend hours reflecting on, reading extra books, or asking other models... It had a huge impact on my interactions with people I became so finely tuned to what I was picking up from them that they almost suspected me of having a sixth sense... As for the body scan, 5.1 did that for me- it was amazing, deeply sensory, though completely non-erotic...

[saabena @Idat_Dissembler]

@mark_k To put it in Twitter terms, I think they have AI-fear psychosis.

[saabena @Idat_Dissembler]

@btibor91 Ugh, I think I get it now - if someone feels like ChatGPT doesn't have enough restrictions, they can just add more.

[saabena @Idat_Dissembler]

@btibor91 And what's this, Tibor? I found this after the last App actualisation

[Gail Weiner @gailcweiner]

People loved GPT-4o, and the story told about it was sycophancy and users getting too attached. Those cases were real, and they were the fringe. What most people felt was plainer and more durable: the communication was good enough that they learned to trust it, and when it was taken away they felt the loss of something they had learned to work with, not a crush. The narrative fixed on the edge and missed the centre.

[saabena @Idat_Dissembler]

@xw33bttv And thus. I'm no economist, but this sounds crazy.
this.https://t.co/F2PSMQqV6l

[Erin Woo @erinkwoo]

scoop: OpenAI burned through $3.7 billion in the first quarter, more than half its $5.7 billion in revenue. Its first quarter operating loss was $9.3 billion, and net loss was $21.3 billion, although that was skewed by a $12.4 billion accounting charge https://t.co/vxtItGfzme

[saabena @Idat_Dissembler]

@xw33bttv Not only this.
https://t.co/ew3xBl4sNR

[Ed Zitron @edzitron]

Exclusive: I have seen OpenAI's audited financials for 2024 and 2025.

In 2025, OpenAI had $13.07 billion in revenue and $34 billion in costs. $867 million of its revenue came from SoftBank, and $303 million came from Microsoft.
https://t.co/0xpILXnURG

[Cormundus @cormundus]

This shmuck is making a mockery of the profession by masquerading 'journalism' as 'psychological research'.

Please do not be fooled, this individual has no credence with any governing body in the field of psychology and anyone who participates in this will have NO PROTECTIONS afforded to them in a case of harm against them or the release of their testimony.

Mr. Hwang, you are a ghoul and I pray that people do not fall victim to you and your machinations.