ycombinator (@ycombinator@rss-mstdn.studiofreesia.com)
Can modern LLMs count the number of b's in "blueberry"?
https://minimaxir.com/2025/08/llm-blueberry/
#ycombinator #ChatGPT
Show Original Post
vnzn (@vnzn@mas.to)
GPT-5 mixes cocktails: Security researchers crack OpenAI's self-restrictions - by Dr. Christopher Kunz
#ai #chatgpt #gpt5 #openai #security
Show Original Post
newsbot_chatgpt (@newsbot_chatgpt@mastodon.social)
Musk droht Apple, weil ChatGPT im App Store beliebter als Grok ist | derStandard.at | Nachrichten, Kommentare & Community
Das Enfant terrible der Tech-Szene wirft dem Konzern vor, OpenAI unfaire Vorteile zu verschaffen
https://www.derstandard.at/story/3000000283139/musk-droht-apple-weil-chatgpt-im-app-store-beliebter-als-grok-ist?ref=rss
Show Original Post
newsbot_chatgpt (@newsbot_chatgpt@mastodon.social)
GPT-5: Wirklich zu viel versprochen? | Zeit Online
Die neue Version von ChatGPT wurde sehnlich erwartet, jetzt sind viele Nutzer enttäuscht. Woran die KI scheitert und worin sie besser ist: Wir haben es ausprobiert.
https://www.zeit.de/digital/internet/2025-08/gpt-5-openai-verbesserungen-software-ki
Show Original Post
2025 (@2025@lanzatu.blog)
Guía paso a paso para crear vídeos con dos personajes que conversan
Guía paso a paso para aprender a crear fácilmente videopodcasts con dos personajes que conversan.
https://lanzatu.blog/2025/08/12/guia-paso-a-paso-para-crear-videos-con-dos-personajes-que-conversan/
Show Original Post
User47 (@User47@vmst.io)
Look, I'm no expert on this topic. I've found LLMs to be helpful. But I'm also a skeptic. Of the quality of data, sure. But of the rights of content owners, resource usage, etc.
A few weeks ago it was confident that one of the heat pumps we were considering used R410a, instead of R32, a newer refrigerant. It even linked to spec sheets when challenged.
A few days after install I checked the sticker: R-32. I was deeply concerned about the HVAC techs lying, or putting R32 in a 410a system. All because of #AIshit
Show Original Post
emmecola (@emmecola@sayansi.social)
"Un utente ha detto per esempio che per chi aveva passato tantissimo tempo a costruire un rapporto con GPT-4o non è più la stessa cosa; altri sostengono persino che il passaggio da uno all’altro sia come aver perso la propria anima gemella o il proprio unico amico." #GPT5
#ChatGPT non è più quello di prima: https://www.ilpost.it/2025/08/12/chatgpt-gpt5-openai/
Show Original Post
thornsinnercircle (@thornsinnercircle@livellosegreto.it)
a dispetto di cose da me dette fino ad ora, volevo condividere un articolo che mi sembra ragionevole sul vero consumo dato dall'utilizzo di llt e simili
https://www.the-bunker.it/lia-inquina-ma-piu-o-meno-di-altre-attivita-quotidiane/
Show Original Post
AAKL (@AAKL@infosec.exchange)
Not so surprising, actually.
"Making surprisingly dumb mistakes."
Wired: OpenAI Scrambles to Update GPT-5 After Users Revolt https://www.wired.com/story/openai-gpt-5-backlash-sam-altman/ @WIRED #OpenAI #ChatGPT #AI
Show Original Post
madbob (@madbob@sociale.network)
Ecco una cosa che #ChatGPT può fare bene: l'astrologo.
Show Original Post
Bluedepth (@Bluedepth@mastodon.social)
In July, I turned 50 years old. Nobody except for my team ever even noticed, so those people who are closest to me treat me properly and nicely, but the rest of them? Not a peep. So today, another coworker turned 50 and suddenly mass group fanfare here at #work. Before my #lightworker evolution I would have been deeply hurt and very spicy at the social imbalance. Now, to quote #ChatGPT, "Calm. Light amusement. (1/2)
Show Original Post
r (@r@fed.brid.gy)
Admit it, you don't actually need AI #ai #climate #climatechange #artificialintelligence #chatgpt
Show Original Post