OpenAI has developed an artificial intelligence that is too powerful to be released
OpenAI is a non-profit research organization on Artificial Intelligence that wants to promote and develop friendly intelligence in the way that humanity can benefit from it. It is founded by Elon Musk in 2015 with the aim of "collaborating freely" with other institutions and researchers making its patents and research open to the public.
One of the projects that the company is carrying out during this period is the one concerning an Artificial Intelligence designed to write texts like human beings. This new AI has been called the GPT-2 and developmental researchers claim to have created something so smart that it is dangerous if released entirely to the public.
There are already Artificial Intelligences that deal with the writing of texts, but we find them especially in sports, where the text is often similar and therefore easier to write for a "machine". In this case, however, we have an Intelligence that has trained through eight million web pages, more than enough to learn the different writing styles of human beings and concentrate them in a single text.
The power of this AI could be misused if made totally public, only part of the code was published to allow other researchers to have a solid base from which to start. In fact, the company said: "Because of our concerns about the malicious use of technology, we will not release the trained AI version." As part of our experiment that aims to divulge the AI models responsibly, we will release a reduced model for the researchers, to allow them to experiment and test the technology ".
The wrong use of this new Artificial Intelligence can lead to the drafting of fake news, automate the "production" of SPAM, automate the production of abusive content on social media or even impersonate people online. The purpose for which this Artificial Intelligence has been realized is different: it can be very useful to improve the writing assistants (find errors and suggest more suitable corrections), to refine a better dialogue with a human being (Google Duplex), to limit errors during text translation and refine speech recognition.
A new and interesting world that of Artificial Intelligence that can also be risky if used in the wrong way, but useful and effective if used appropriately to improve our experience in the technological field. We must always keep an eye on when we surf online; even OpenAI warns us to be more attentive to the content found on the web, as it will be increasingly difficult to distinguish the true and false.