Artificial intelligence (AI) does  not stop advancing. The American company  OpenAI  announced on Tuesday the long-awaited launch of  GPT-4 , the latest generation of a much more powerful language model that “exceeds  ChatGPT  —its popular chatbot— in its advanced reasoning capabilities,” thus getting closer and closer. to simulate a human conversation.

In a statement on its website, the research company explained that GPT-4 is more reliable, creative and capable of handling much more nuanced instructions.” In addition, the new model will be able to generate new content with both text and images. For now, this feature will only be available to paid subscribers of the platform.

The popularization of  AI  points to the transformation of a multitude of professions. And it is that tools such as those promoted by OpenAI allow users to automate all kinds of tasks, a possibility that is already affecting schools and universities. ChatGPT, for example, is capable of answering all kinds of questions, summarizing complex texts, writing novels imitating specific literary styles, composing poems, making jokes, and even programming computer code. All this almost in real time. Others, like  Dall-E  or  Midjourney , allow images never seen before to be created from a simple description, changing the conception of what we understand by  art .

More accurate but still failing

However, the operation of these tools is far from perfect. ChatGPT uses GPT-3, a very advanced model but still incurs errors. In recent weeks, AI experts have detected that this robot can respond to users with incorrect or false information, presenting it as true and replicating biases. These imprecise responses are known as “hallucinations.”

GPT-4 is more accurate than its predecessor. Thus, it would obtain a score 40% higher in certain veracity tests and would be 82% less likely to respond to requests for disallowed content, according to OpenAI. Sam Altman, executive director of the fashion company in the sector, has assured that this latest model is “more capable” but that “it continues to have flaws.” A warning to distrust the product that they have launched on the market.

business battle

AI is destined to be the  technology  of the next decade and to alter the way the  internet works . That promise is what has launched large companies in the sector to compete to lead an emerging business that already moves billions of dollars around the world. Microsoft  owns almost half of OpenAI after investing $10 billion in the startup in January.

Earlier this year, the tech giant announced that it would integrate a conversational chatbot feature similar to ChatGPT into its  Bing  search engine , potentially transforming the way we search for information online. Now, GPT-4 will serve to further improve how Bing works. Microsoft has also started this integration of AI in other products such as Word, Excel, the Edge browser or its Azure cloud services.

Google has been a powerhouse in the development of AI for years – it invented an essential part of the GPT model in 2018 – however until now it had chosen to keep its innovations under lock and key for fear that its negative effects would harm its reputation. Microsoft has chosen the other way, which has allowed it to control the story and position itself as a market leader. That strategic gesture has forced  Google  to react. In January it announced the  launch of  Bard , its own text generator . In the past few weeks, it said it would release new AI features for Gmail, and just yesterday it said it would allow developers to use its language model.

LEAVE A REPLY

Please enter your comment!
Please enter your name here