As you all probably know by now, AIs have been on the rise lately, and if they can make something as complex as art, you can bet they’ve come a long way in literature as well. Huh. What better proof of the above can there be chatgpt A revolutionary artificial intelligence prototype developed by OpenAIOne of the leading companies in its respective field and backed by Microsoft.
Since its launch in late November 2022, ChatGPT is causing a sensation for its unprecedented level of accuracy, Being able to have believable conversations with human interlocutors, giving correct answers with natural language, writing texts on practically any topic, creating stories, writing newspaper articles, translating multiple languages and even programming languages like Python Writing usable code in .
ChatGPT system is based on deep learning contains over 175 million parameters and is trained with millions of books, so it has access to an extremely large database.
Also, use an autoregressive language model, which means that your answers are based on previous comments. These observations can be based on other points in the conversation, creating a sense of context with your interlocutor.
Naturally, the interactions and tests users give to the system serve to improve it and enhance its capabilities.
How to use ChatGPT
for now ChatGPT is free to use And technically anyone can access it by visiting the official site chat.openai.com, where we will normally be asked to register and verify our account. Once there we can ask any questions or make requests to ChatGPT, who also understands Spanish.
The site explains some examples of AI’s capabilities and limitations, including the fact that can sometimes generate wrong information, harmful or biased on any subject. Likewise, we are reminded that your knowledge of historical facts and world events is limited to events after 2021.
It should be noted that Currently not available in all countries And its surge in popularity has resulted in high traffic at times in which the service is not operational.
As we mentioned earlier, OpenAI is transparent with the limits of its AI model, but an interview with carlos sierraThings go further and it can be said that AI applications “mimic, but do not understand”, Director of Artificial Intelligence at CSIC and President of the European Association of Artificial Intelligence.
what are your limits?
In this sequence of thoughts, Sierra explained that in fact behind all this is a machine “incapable of understanding what is true and what is not.” Hence, you may end up making a lot of inaccuracies when dealing with complex issues. It, along with other similar systems, is capable of generating grammatically and orthographically correct sentences; which create the illusion that they were written by a human author, but the truth of the matter is Correct combination of phrases already established in the training set,
The system basically looks for historical input and uses a sequence of similar phrases, resulting in an output phrase. “It is like mixing up all the history books, but I insist that all these productive systems do not understand what they are doing. They limit themselves to copying, but do not really understand .
Similarly, the president of the European Association for Artificial Intelligence stressed that more inconsistencies can be observed as the writing moves away from the first sentences. This is particularly notable in long texts and is due to this loss of coherence in normal discourse that can be understood between text generated by artificial intelligence and text written by a real person.
All that said, and despite concerns about the potential consequences of this technique, it should be accepted it’s pretty impressive what you can already do And that it has great potential in the educational and technical fields.