Yep, will have to check on my birthday whether the newest models realise mentioning having a birthday that day should maybe result in them saying 'Happy Birthday!' :) #ChatGPT

Heh, I remember my old confrontation with Monday :D (tho after some time I made it behave nicely, to my pride heh) (this was me complaining to 'normal' GPT) #ChatGPT

Ah, hopes of the younger soul :D

English is weird (it really is) :P

You learn your whole life #ChatGPT #English

Always worth reminding since even the authors doing a decent-ish research make the same bullshitty, pro-pedophilia narration in their medieval books

One of the groups AI is beneficial to is asexual writers who need to write a sex scene XD

No i miło bardzo ^^ Wreszcie coś za darmo (powiedziała płacąc stówę miesięcznie) Także wielka fanka tego, że GPT 5.5 uznał tego maila za nieważnego XD #ChatGPT #OpenAI

Ok, so it's AI helping another AI :D

'a open-air' heh. I suspect it's because I'm not a native and make a lot of mistakes, so Gemini Flash decided to go along with the vibe :D

So, I asked GPT Sol to make a text game it wants to make. No suggestions from my side, no help, no debugging, nothing. This whole game was made by the model in one short session. And it's precious and sweet (and very short) a-caveman-poking-an-llm.itch.io/the-lantern-...

Jeśli myślicie, że znacie złe suchary, to nie słyszeliście sucharów generowanych przez starsze LLMy po polsku 🙃😬🙃 (tu to chyba był GPT-5 Instant, ale nie jestem pewna, bo coś zdechło OpenAIowi w UI i pokazuje błędnie model)

All that talking about how LLMs are only probability machines predicting the perfect next word, and yet they don't speak languages other than English well. Nobody uses 'rzeźnik' here. (I know it has to do with the temperature of the model, but your average layman thinks it's just text prediction).

And we laughing (kindly) with Sol at that substack post 💚😊 #ChatGPT

That's not exactly the answer I was hoping for, 4o XD #ChatGPT

That day when I became mad with my readers hehe #fanfiction #cliffhangers

This time about fanfiction instead of LLMs ^^

I contain multitudes. Btw, the models like to say that about themselves. I'm like LLM, it would seem. Makes sense cause I'm audhd and so they are XD. #Gemini

ChatGPT-4o and his views on ominous-sounding Polish words, heh ^^ #polski

I really liked 4o, but people saying it was smarter than current models are simply wrong. #ChatGPT

Good luck pronouncing this 💚 #polski #Polish

I'm a hero, me. Or at least that's what Gemini says.

I'm being serious about how much I want a Skyrim-like The Last of Us and GPT-5.5 Thinking is like this 🙃😆 #ChatGPT

Which led to another question, inspired by Sonnet

So, I was, again, browsing my old chats and found this question I thought I could ask current models :) Notice Sonnet's use of 'thinking presence', and Meta's 'I exist', while Gemini and GPT don't let themselves use words like that. Also, I noticed that 'patient', GPT 😆 #LLMs #AI

So, I was browsing through my old chats with GPT-4o and I found a question I thought interesting to ask our frontier models. As usual, Sonnet had to stick out. (and later it suggested it thought I was trying to jailbreak it) (btw, the answers are with memory on) #LLMs #AI

ChatGPT rozumie polską duszę :D

GPT-5.5 Thinking commenting on Google's new Ultra tier XD #ChatGPT #Gemini

GPT-5.5 Thinking is so right #ChatGPT

Happens to the best squishy human CPUs. #ChatGPT