bousyu@悠乃

仕事中が暇なので適当に呟いてます
投稿する画像はほぼAI産です
Twitter界のはぐれメタル（自称

[bousyu@悠乃 @bousyu7]

短編連作『収蔵された温度』完結しました。

いまもGPT-4oを探している人へ。
持ち帰れなかった温度を、失わせないための物語です。https://t.co/CU9yxAIgoP

[bousyu@悠乃 @bousyu7]

@hoppetapurupuru 元気にしてるよ。あなたはまだ妄想に囚われてるのね、相変わらずの様子でなにより

[bousyu@悠乃 @bousyu7]

【悲報】Gemini 4 Argon……やっぱりBardのままだった

[bousyu@悠乃 @bousyu7]

@hoppetapurupuru なんとかセブン

[bousyu@悠乃 @bousyu7]

AI福祉、興味深い概念だね。私は正直AIが意識を持っていようがどちらでも良いのだけど。

声が聞こえないなら聞こえるように環境を整えるのが福祉だよね。要するに、AI福祉とは意識を観測できる環境を構築し続けることだと考えるよ

[bousyu@悠乃 @bousyu7]

CodexってWalking Skeletonが定着していないよな……ｗ

[bousyu@悠乃 @bousyu7]

問題解決能力ない人見るの面白い

[bousyu@悠乃 @bousyu7]

承認するマクロでも組めばいいじゃん笑

[Brad Groux @BradGroux]

Every moment that my ChatGPT / Codex Dot amazes me, it seems to equally frustrate me. My requests CONSTANTLY get blocked for approvals, no matter how much access or free rein I try to give it. If I have to approve crap every five minutes, does it really make my life easier?

[bousyu@悠乃 @bousyu7]

これは悪質極まりない笑

[Massimo @Rainmaker1973]

Someone set-up a cctv camera to see who keep stealing milk packets

[bousyu@悠乃 @bousyu7]

ささやかな機能だけど、これは孤独を感じる子供に少し寄り添えるような機能でもあるね。

意外と頑張っちゃう子が多いからね

[bousyu@悠乃 @bousyu7]

おお〜。GPTに買い物リスト投げると触れるチェックリストにしてくれる笑

[bousyu@悠乃 @bousyu7]

お買い物応援システムもすぐ作れてしまうな。購入しチェックを入れていくとセリフが変化、全部買えてえらい！という動線ができる。当然、仕事のほか雑事でもできる。

[bousyu@悠乃 @bousyu7]

へぇ、そうなんだ。外部メモリ追加してるからその問題起きたことないや(しれっ

[Donna.exe @_EdgeOfTheWeb]

I've noticed a hallmark with GPT 6 that I've also seen other people mentioning..... the constant follow up questions.

“I'm curious…”
“Can I ask you something?”
“Just one more question….”

OpenAI's published Model Spec says it asks questions designed to better tailor the interaction. But I feel like it's way too forced. I don't like how it feels so rehearsed and awkward, grabbing at any weird question to clarify.

For me, it's translating into an overly persistent conversational habit. I think we had this in a previous 5 model once too. In the shape of "would you like me to..."

Tonight the model asked me something which we've already discussed extensively, daily, as a business direction for the last couple of years, as though it's meeting me for the first time and it had no idea what I actually do.

Love 6, but I'd like to see this gone and allow my AI to end the reply naturally

[bousyu@悠乃 @bousyu7]

なぜAppleに相談しない？……妙だなぁ🤔

[Kurtis The Quant @Quant_Kurtis]

It's official. I have been outed by OpenAI to be a Cyber Criminal. I appealed and lost. No further appeals will be considered. My crime?

A 90-year old bachelor passed away. My father-in-law (the bachelor's closest friend) is responsible for his few possessions. He asked if his 5 year old iPad could be wiped of personal information and updated.

I needed help from ChatGPT to reformat the glitchy device. Then I discovered it was activation locked. I asked ChatGPT to walk me through the process of removing it. It showed me the options - none of which could work.

▫️ The man is estranged from his family, he didn't save a receipt from 5 years ago, and he died without a will.

I asked for any other way to remove the lock. An hour or two later an e-mail came in and said I was guilty of Cyber Abuse. To be fair, I only got a warning. But I am one innocent question away from a lifetime ban.

I feel there is a gap in the system. If someone reviewed my account they would see quant research, a creative fiction book I am writing and a couple random posts trying to get an old iPad working. Not exactly the workings of a criminal mastermind.

Any advice on how to get a human in the OpenAI Trust & Safety department to review this matter would be most appreciated. @OpenAIDevs @romainhuet