What is ChatGPT? “I am a language model, which means that my main function is to generate text in a coherent and fluent way in response to questions or indications given,” explains the assistant himself, (assistant), as he likes to be called.
Launched on November 30, 2022 for free by the American company OpenAI, ChatGPT is a computer program system or “chatbot” with a language based on artificial intelligence, that is, the creation and design of elements capable of reasoning on their own. themselves using human intelligence as a paradigm.
This application, which became a viral phenomenon a few days after it was launched, uses GPT-3.5, a language model with more than 175 million parameters, which is trained with a huge sample of text taken from the Internet to answer to language-related tasks in real time.
WHAT IS IT FOR
Still in the testing phase and therefore with limitations, however, it allows you to create any type of text, summaries, newspaper articles, jokes, scripts, or even create almost à la carte content and you can do it with the desired number of characters and with a speed of response greater than the time taken to formulate the question.
It allows chatting in a multitude of languages, “it is coherent, it takes into account what has been previously said in a conversation and it is capable of identifying topics that it is better not to talk about”, as EFE has verified.
OpenAI itself, the company that created it, in whose foundation the American billionaire Elon Musk participated, but from which he later disassociated himself, highlights the dialogue format with which it operates and the possibilities it generates from this tool.
"The dialog format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests."explains the firm.
HOW DOES IT WORK
“I have been trained using a large amount of text written by humans.” “My goal is to imitate human language as precisely as possible,” says the assistant himself.
However, and although it is trained to read millions of web pages, ChatGPT is not connected to the Internet and its “knowledge” stops in 2021, so for the moment it cannot respond to events that have occurred recently.
It makes mistakes and cannot, for example, forecast the weather, or learn from its mistakes, as human intelligence does. However, if someone questions one of your wrong answers, the system admits its failure and tries to find a solution.
It is precisely the technicians who are currently working to fill this gap in the development of training techniques and the contribution of more data.
ALTERNATIVE TO GOOGLE?
Meanwhile, and although its results are sometimes erroneous or contradictory, there are already those who see ChatGPT as a quick alternative to a Google search and as a counterpart the internet platform has already launched “Bard”, its new "chatbot"also created by AI in apparent response to that of its competitor.
This being the case, and despite the news, questions and limitations, there is no doubt that the future is already here.
According to Sam Altman, CEO of OpenAI, ChatGPT is by now a "early demo" of what will be possible to do with language interfaces based on artificial intelligence.
"Soon you will be able to have helpful assistants who talk to you, answer questions and give advice. Later on you can have something go and do chores for you. Eventually you can have something that discovers you new knowledge"Altman said on Twitter the day of the assistant’s presentation.
THE “BUTS”…
In January 2023, just launched the "chatbot" From OpenAI, Mike Sharples, Emeritus Professor of Educational Technology at the Open University in the United Kingdom and pioneer of the new chat system, defended the critical use of this tool in education, since he said, "allows us to learn from the immense possibilities offered by" although at the same time he warned of the risks that it entails for the "democratize plagiarism"a new "challenge"for academic institutions from now on, along with the risk of students using this system to afford not to produce their papers.
In this sense, Sharples also highlighted that, although this technology "is enormously capable of producing credible texts", "cannot have real-time information, cannot reflect on what they have written, does not have an explicit model of how speech works, and is amoral"given that "is a language machine, not a reasoning system".
This is the reason why the Institute of Political Studies in Paris, one of the main training centers for French political and administrative elites, announced in 2023 the ban on ChatGPT, without explicit mention of it in the written or oral works of its members. students. And the same backlash has come from schools across the United States, which have been quick to limit use of the app out of fear the software could undermine learning.
Despite everything, Professor Sharples himself predicted that the use of AI in the classroom will be similar to that of the first calculators or mobile phones, where the response "first it was ignoring them, then rejecting them, banning them and finally trying to adapt".