Mexico City.- Communicating with an artificial intelligence that responds as if it were another person and seems to know everything is now possible with ChatGPT, the most advanced trained language model ever created by OpenAI.
“This model uses an advanced technology called ‘deep learning’ (deep learning) to process natural language and produce appropriate responses to the questions that are asked.
“It has been trained on a large data set, including millions of conversation and text examples. This allows it to understand the context and intent behind the questions and produce reasonable answers, similar to how a human would.” , explains the program in a conversation through its website, opened on November 30 and which added a million users in five days.
In it you can chat with the model and receive consistent responses.
He also follows social conventions, such as greeting, thanking, and apologizing.
He is capable of finding errors in programming codes, making one and guiding the process of applying it, creating a story or poem and even writing a song.
Regarding its possible applications, it indicates that it could be used to improve the capacity of messaging bots by answering questions in a more precise and natural way, or helping to write longer and more complex messages in applications.
You can also suggest ideas for video games or propose a message tailored to the interests of a “match” made on a dating application or write a short academic essay.
Despite this ability and offering your answers in a few seconds, this tool does not guarantee certainty.
“ChatGPT sometimes writes answers that sound plausible but are incorrect or nonsensical. Solving this problem is challenging, since during RL (reinforcement learning) training, there is currently no source of truth,” warns on its OpenAI page.
So, one of its main risks is that it is used to create false texts, even as news or imitating the discursive style of a transcendent character.
To fine-tune its current security filters and improve the model, it was opened to the public, since feedback is provided with each conversation, which costs the organization less than 9 cents.
#Artificial #intelligence #solve #doubts