Openai announced on Thursday, August 7, the launch of the Chatgpt-5 model, a version that promises better results in tasks such as the response to health questions, writing and generating computer codes.
Photo: Pixabay
According to Washington Post, Openai representatives previously announced that GPT-5 would not perform at this level.
In a demonstration broadcast live on Thursday, the new system has shown that, like the previous models, it can still make mistakes.
When a director asked Chatgpt-5 to explain how the scientific principle called the Bernoulli effect is applied to the wings of the aircraft, he provided an ordinary but incorrect explanation.
All Chatgpt users will have access to the GPT-5, even those who use the free version. But only those who have a subscription “Pro“$ 200 per month will have unlimited access to the newly launched system. The GPT-5 will be the default mode on all versions.
Users who do not pay for Chatgpt will only be able to ask a certain number of questions to which GPT-5 will answer before the chatbot returns to the use of an older version of the OpenAi technology.
GPT-5 answers questions faster than previous OpenAi offers and is less likely to “Halucine“Or invent false responses, Openai directors said in a press conference before launch.
The new Openai’s new software can answer questions using a process called reasoning, which shows the user a series of messages trying to break down a question in steps before giving the final answer. “GPT-5 is the first time you really feel like talking to an expert, an expert at doctoral level“Said Sam Altman, CEO of Openai.
Altman said that the GPT-5 is particularly good in generating programming codes, a feature that has become an important selling argument for Openai and the developers of Rivali and has transformed the work of the programmers.
In a demonstration, the company showed how two instruction paragraphs were sufficient for GPT-5 to create a simple website that offers meditations in French, complemented with a game and daily vocabulary tests.
Altman predicted that people without training in computer science will one day generate quickly and easily any kind of software they need to help them at work or other tasks.
“This software idea on request will be a defining part of the new GPT-5 eraAltman said.
In a live broadcast on Thursday, OpenAI directors said that ChatGPT users can now connect the application to Google calendars and email accounts, allowing the chatbot to help people program according to existing plans.
The Openai CEO, Altman, was one of the greatest supporters of the idea that capabilities grow so quickly that many aspects of society will soon revolutionize.
The changes made by Openai to Chatgpt in April triggered negative reactions online after examples of chatbot that seemed to flatter or manipulate users went viral. Openai said he worked to reduce this trend in GPT-5.
Openai on Thursday published a report on the capabilities and limits of GPT-5, mentioning that the company carefully analyzed the risks of psychosocial damage and collaborated with Microsoft to test the new IA system. The report mentions that the GPT-5 reasoning version could still be “Improved in terms of detection and response to certain specific situations in which a person seems to face mental or emotional suffering.”
Earlier this week, Openai said in a blog post that he collaborates with doctors from over 30 countries, including psychiatrists and pediatricians, to improve how Chatgpt responds to people in times of suffering. Turley, the director of Chatgpt, said the company does not optimize Chatgpt for interaction.
