Published on
•Updated
ADVERTISEMENT
Openai will launch, Thursday August 14, the highly anticipated GPT-5, the latest model of its popular artificial intelligence chatbot (IA) Chatgpt.
This new model has advanced writing capacities, invents more often answers when it does not have good information to answer a request, and better answers health questions by signaling potential medical problems, depending on the company.
GPT-5 has capacity “agentics “ that he will use to develop code, which means that he will be able to perform certain tasks and make decisions without human intervention.
The updated model can create websites, applications and games, and can be integrated into user Google tools, such as Gmail and Google Calendar, said the Openai company.
Users of a preliminary chatgpt version will also be able to choose between four “Personalities“For the chatbot.
IA enthusiasts saw this drop after the publication, in the early hours of Thursday morning, of a description of the update on the Crowdsource Github coding platform, which has since been deleted.
The launch of GPT-5 follows the introduction by Openai of the Open-Source GPT-OS series earlier this week. This is the first model for which the company has also published the source code since the 2019 GPT-2, so that users can refine it to meet their specific needs.
Why is this launch important?
The launch is the latest development on the path of agentic AI, where agents can carry out actions in several steps and use tools to navigate the web without the help of human prompts.
According to an example given by Amazon, an AI agent in a customer service center could thus automatically ask questions, search for information in internal documents, answer by offering a solution or send the customer to a human if necessary.
At the beginning of the year, the CEO of Openai, Sam Altman predicted that the first agents of IA were soon going “Join the active population” And that they were going to “materially change the production of companies”.
In a blog post published in June, the CEO of Open AI took the example of an AI agent in a software engineer role, saying that the agent “will eventually be able to do most of the things that a software engineer of a large company with a few years of experience could do, for tasks of a maximum duration of a few days”.
“He will not have the greatest new ideas, he will require a lot of supervision and human direction, and he will be excellent in certain things and bad in otherss, “he wrote.
Agents could also be considered virtual colleagues “reELS but relatively young “ which could be adapted to “All areas of knowledge work“He continued.
The launch of agent agents by Openai would not be the first of a company specializing in AI. Microsoft and Google have already launched AI agents which can be personalized according to the services that a company must provide or the tasks it wishes to automate.
