According to TechRadar , OpenAI has just officially launched GPT-4.5, the latest version of the AI (artificial intelligence) language model on ChatGPT, promising to bring more natural conversations and 'understand' users than ever.
GPT-4.5 improves on 'emotion'
The highlight of GPT-4.5 is its ability to significantly improve emotional intelligence (EQ). Although AI cannot actually feel, the model is designed to recognize nuance, understand context, and respond more empathetically. Instead of robotic responses, GPT-4.5 aims to make it feel like you are talking to a real friend.
OpenAI officially announces GPT-4.5 model
PHOTO: THE VERGE SCREENSHOT
OpenAI engineer Rapha Gontijo Lopes shared that the model's EQ is assessed through emotion, which is the level of cooperation and warmth in the tone of voice. To measure this, OpenAI had experts evaluate and select the most appropriate responses.
Additionally, GPT-4.5 is also improved to reduce the problem of 'illusion', that is, confidently giving false information. This is a common problem in previous AI models and a concern for users.
In terms of performance, GPT-4.5 is highly regarded for its ability to process general knowledge. However, OpenAI also notes that this model is not as specialized in programming and complex problem solving as some other AI models.
GPT-4.5 is currently available as a research preview for ChatGPT Pro users (which costs a hefty $200/month). ChatGPT Plus and Team users will have access to the model next week. OpenAI hopes to get feedback from users to continue improving the model.
The launch of GPT-4.5 comes amid an increasingly intense AI race. Amazon recently announced its AI-powered Alexa+, demonstrating that tech companies are working to develop chatbots that can interact more human-like.
Source: https://thanhnien.vn/openai-ra-mat-gpt-45-tang-cuong-tri-tue-cam-xuc-18525022817344272.htm
Comment (0)