GPT-4 is coming, the new version of ChatGPT artificial intelligence

although of ChatGPT being talked about recently, the reality is that this technology was created a few years ago and the tool that makes it possible, GPT-3, He was born in 2020, so after his growth a new version is expected.

At an event in Germany, engineers from Microsoft They confirmed that the update will become a reality in the third week of March, with a great package of news, starting with a higher quality in their responses.

The American company is involved in Google Tags after your investment in Open AIthe company that owns the technology, and after its integration into the Bing browser, which now has a chatbot, and which has represented a recent increase in user numbers, exceeding 100 million daily active accounts.

It may interest you: How to create a photo studio with artificial intelligence

This technology is the one behind the operation of ChatGPT to produce answers, texts and all the content that marked the trend of this artificial intelligence (IA).

However, it has limitations due to the capabilities the development engineers had at the time, like, for example, it doesn’t have data beyond 2021 and for that reason some answers that involve events after that date receive an incorrect answer. , like not knowing who won the 2022 World Cup.

You may be interested: How to choose the length of ChatGPT replies

In view of its evolution, a new version is essential and it is the one that will arrive soon, promoted by Microsoft, which is about to launch a system called Kosmos-1 which consists of a large multimodal language model (MLLM) this means it is an AI capable of receiving input via text, images , voice or videos, which gives him more possibilities to understand the requests of the user and to be more precise in the results.

For that, GPT-4 it will have many more parameters than its previous version; that is, with more data than necessary to generate their learning. And the increase would be 175 trillion for the current version, to 100 trillion for its update.

This will allow you to respond to more complex situations and generate even more “human” responses, according to the engineers of Microsoft.

It may interest you: Artificial intelligence applied to mental health: lack of control can have serious consequences

An example of all this is that this technology would be able to receive inputs through video, this way it can learn from the images shown, the person’s movements and feelings, the audio, the voice tones and the fluency of the conversation to extract this information and generate content, which can be a summary of a film or a lecture.

Microsoft engineers gave another practical example in the world of work. They talk about the importance the new version would have in areas like customer service, where technology could transcribe conversations and summarize them, optimize the time a worker takes, save 500 hours of work a day by receiving 30,000 calls per day. .

“It’s not about replacing jobs, but about performing repetitive tasks in a different way than before,” said Andreas Braun, CTO of Microsoft Germany.

It may interest you: They thought their loved ones were asking them for help: it was a scam with artificial intelligence

Along with all this, from the company, they assure that artificial intelligence will continue to make mistakes and that it is essential to validate their information. Likewise, Sam Altman, CEO of OpenAI, clarified that it’s important to balance expectations with this breakthrough, because “people are screaming about being let down, and that’s what’s going to happen.”