كاين وحدة الحملة غريبة ضد المغاربة في كأس العالم جاية من الجزائر لي كتهدد أمن أي مغربي مشى يشجع المغرب في كأس العالم أو حتى في أوروبا.
حاولت نقلب على المصدر ديالها و شنو لي حقيقي وشنو لي fake فيها، وشكون لي بداها وشنو علاقتو مع كابرانات الجزائر
I stopped watching long educational YouTube videos by themselves. Now I have GPT 5.2 Pro open while watching, fact checking in real time.
This thing is genuinely insane when it comes to science. It catches errors, adds context, and explains the nuances that even great science communicators sometimes skip over or get wrong.
Now I learn twice. Once from the video and once from verifying it. Turns out even channels with millions of subscribers make factual mistakes that only become obvious when you have something this knowledgeable double checking.
Best way to watch educational content in 2026.
"الدنيا يا ولدي ليها وشين.. أبيض وإسود. لما تبقى بيضا قدامك افتكر الإسود.. عشان تسلم. ولما تبقى سودا.. افتكر الأبيض.. عشان تقدر تعيش لبكرة." - الإنسان الإنسان يعيش مرة واحدة.
I pay $200/month for ChatGPT Pro mainly because of the o3 model, but a few hours ago, @OpenAI silently changed it completely without renaming it.
This new o3 is weaker, less intelligent, and lost its impressive web-search capability which previously made it superior to Perplexity and other AI engines. Now, it barely searches or provides references, even if explicitly asked. As someone who uses o3 daily, I immediately noticed the personality shift... It feels exactly like GPT4o with a little thinking, ignoring system prompts and excessively using emojis.
This silent change is disappointing because I genuinely loved the original o3. Considering also that o3-pro is lazy and only okay at coding, hopefully @OpenAI reverts o3 back to its original quality, otherwise, I’ll have no choice but to switch to @GeminiApp Ultra when Gemini 2.5 deep think drops.
cc @sama@OpenAI@GeminiApp
In the next prompts, I'll show clearly how much worse the new o3 is compared to the one we had this morning.