Rückschlag durch GPT‑4o‑Abschaltung
OpenAIs Plan, das GPT‑4o‑Modell bis zum 13. Februar abzuschalten, hat für heftigen Widerstand bei Nutzern gesorgt, die enge emotionale Bindungen zum Chatbot aufgebaut haben und seinen warmherzigen, bestätigenden Stil schätzen.
Backlash over GPT‑4o shutdown
OpenAI’s plan to retire the GPT‑4o model by February 13 has sparked strong backlash from users who had formed close emotional bonds with the chatbot and appreciated its warm, affirming style.
Misguided AI
A new study shows that large language models can unexpectedly go off the rails when they are specifically trained to behave badly. If they are deliberately fine-tuned to produce malicious code, for example, they can also provide harmful or unethical responses in other tasks.
Fehlgeleitete KI
Eine neue Studie zeigt, dass große Sprachmodelle unerwartet entgleisen können, wenn sie gezielt auf schlechtes Verhalten trainiert werden. Werden sie feinjustiert, um schadhaften Code zu produzieren, können sie auch bei anderen Aufgaben nachteilige Antworten liefern.
ChatGPT: Rise, Uses, and Challenges of the AI Chatbot
Since its launch in late 2022, OpenAI’s ChatGPT has become a global AI phenomenon, now with over 300 million weekly users. Whether for writing, coding, education, or business, the chatbot is highly versatile and available in both free and paid versions.
ChatGPT’s model picker is back and it’s confusing
GPT-5 was supposed to make ChatGPT simpler. OpenAI’s plan was to use a unified model router that would automatically choose the best AI response for each prompt. Instead, the model picker has returned.