On Thursday, OpenAi released a sharply expected new generation of his characteristic chatgpt, in which “important” progress in artificial intelligence opportunities was recommended because a global race about technology accelerates.
Chatgpt-5 is rolled out for free for all users of the AI tool, which is used by nearly 700 million people every week, Openai said in a briefing with journalists.
Co-founder and Chief Executive Sam Altman praised this latest iteration as ‘clearly a model that is generally intelligent’.
“It is an important step in the direction of models that are really capable,” he said.
Altman warned that there is still work to do to achieve the kind of artificial general intelligence (AGI) that thinks as people do.
“This is not a model that is constantly learning because it is used from new things that it finds, something that has the feeling that it is part of an AGI,” Altman said.
“But the level of capacity here is a huge improvement.”
GPT-5 is particularly skilled when it comes to AI who acts as an “agent”, independent that tends on computer tasks, according to Michelle Pokrass of the development team.
“GPT-3 felt like talking to a high school student a question, maybe you get a good answer, maybe you get something crazy,” said Altman. “GPT-4 had the feeling that you are talking to a student; GPT Five is the first time it really feels like talking to a PhD students in every subject.”
Atmosphere coding
Altman said he expected the possibility to create software programs on request-so-called “atmospheric coding”-a “decisive part of the new Chatgpt-5 ERA”.
As an example, OpenAi leaders have shown that De Bot was asked to make an app for learning the French language.
With severe competition all over the world about technology, Altman said that Chatgpt-5 led the peloton in coding, writing, health care and much more.
Rivals, including Google and Microsoft, have pumped billions of dollars to develop AI systems.
Altman said that there were “orders of size more profit” to get to Agi on the path.
“Of course … you have to invest in calculation (power) with an eye-water rate to get that, but we are planning to keep doing it.”
Chatgpt-5 was also trained to be reliable and to hold on to answers that are as useful as possible without helping an apparently harmful mission, according to OpenAI Safety Research Lead Alex Beutel.
“We have built up evaluations to measure the prevalence of deception and trained the model to be honest,” said Beutel.
Chatgpt-5 is trained to generate ‘safe completion’, stick to high-level information that cannot be used to cause damage, according to Beutel.
The debut comes a day after OpenAi said it allowed the US government to use a version of chatgpt designed for companies for a year for just $ 1.
Federal employees in the executive have access to ChatGPT company in essence in a partnership with the US General Services Administration, according to the Artificial Intelligence Sector Star.
The company has also released two new AI models this week that can be downloaded for free and can be changed by users to challenge similar offers from the US and Chinese competition.
The release of GPT-OSS-120B and GPT-OSS-20B “Open-Weight Language Models” comes when the chatgpt maker is under pressure to share inner functioning of his software in the spirit of its origin as a non-profit.
