I have already commented in a post that OpenAI is an artificial intelligence development company that, according to it, has the mission of making sure that General Artificial Intelligence is respectful of the human being.
Among other things, they are developing a natural language processing system (NPL) capable, for example, of writing a coherent full story by giving it only one opening sentence, which they have called the Generative Pretrained Transformer (GPT).
Last year they completed the GPT-2 version and said that it could not be released because it could be too dangerous for humanity but. obviously they rleased it out just anyway and realized that humanity had not been destroyed so they continued working on the project.
The first version of GPT was created in 2018 and 110 million parameters were used for its training, last year for the GPT-2 version 1.5 billion parameters were used and for this GPT-3 version 175 billion parameters were used.
With all this, the algorithm is capable, for example, of preparing an article on any topic that is proposed in a matter of seconds, taking care of grammar and style, in addition to maintaining coherence from paragraph to paragraph.
But the thing goes further because, although the algorithm is not trained for it, if we give it a list of sums all solved except the last one, the algorithm will give us the answer to that sum using only the language since it has not been trained to add.
Itcould even be trained to generate computer code from natural language or translate a program from one programming language to another, in fact Microsoft has already presented a product to be able to translate python, java or C ++ programs from one to another without errors.
What does all this mean ?, because all those who thought that robots would not take our work away from us, we are no longer free, the new chatbot created with this algorithm will be able to carry on a conversation on any topic in a coherent way.
Versión en español