OpenAI ad GPT-4, its next-generation model of language IA
The artificial intelligence to advance, and one of the stars in these last few months was ChatGPT, who is not to evolve. OpenAI, the creator of ChatGPT revealed that GPT-4 will be able to accept text and images.
According to the statements of the company, this new model is “more creative and collaborative than ever” and “is capable of solving hard problems with great precision.” In turn, it now supports both the possibility of writing or uploading images, even though the response on the part of GPT-4 will be by text.
GPT-4: the new model
This new model has been trained with large amounts of data through datasetsto learn and generate a language similar to that used us, the humans.
OpenAI warned that it still has some inherent problems of a model of language that is “new”, which includes a tendency to invent information or “freaking out”, and the capacity to generate texts with violent and harmful. Beyond all else, is a huge step forward for AI. According to the company passed 6 months doing that GPT-4 to be more secure and “aligned”. Getting to have a “82% less likely to respond to content requests is not allowed, and a 40% more likely to produce objective responses that GPT-3.5 (the predecessor)”.
For those who are not aware, GPT-4 is the latest model of language that just released yesterday (14/03/2023) OpenAI, which drives applications such as ChatGPT and the new Bing.
Artificial intelligence: GPT-4
Obviously big companies have not been left standing, since OpenAI has stated that it has partnered with a number of companies to integrate GPT-4 in their products, as is the case of Duolingo, Stripe or Khan Academy, for example. For the time being, can only be accessed by those who have ChatGPT Plus, the monthly subscription of OpenAI. But in addition is driving the chatbot Bing (Microsoft).
The CEP of OpenAI, Sam Altam, tweeted that GPT-4 “that remains to be defective, remains limited”, but that also “still looks more impressive in the first use after spending more time with him”. Has also stated that “people are going to disappoint you and always will be, the hype is so... we do Not have a AGI real and that's what they expect of us.”
However, it is still completely revolutionary, and we know that expectations are very high. If you have been diving for internet in the last few months have seen the amount of things that is able to do ChatGPT, from compose songs, write blogs, or stories. Now that is potentiated, because “you can generate, edit and iterate with users in tasks of creative writing and technical, as composing songs ,writing scripts, or learn the writing style of a user”.
One of the most important features is its essence multi-modal, that is to say the possibility of accepting the input of text and images simultaneously,which allows the model interpreter for more complex things.
As we said above, is a major breakthrough for Artificial Intelligence, and more and more companies are incorporating this type of technology to their products. Us Spicytool we are already working for that GPT-4 enhances our tool Google Ads to get better results.
