ChatGPT: what’s the AI ​​program everyone’s talking about?

What is it about?

ChatGPT is a natural language processing software created by the American organization OpenAI that generates text automatically. This artificial intelligence program was built on the principle of “machine learning”, which consists, in part, of relying on a large database, consisting mainly of documents obtained from the web – but not the entire web, which eliminates violence, racism, sexist content, etc. The ChatGPT program (more precisely ChatGPT-3, for the latest version that makes media news) can be used in several languages, including French.

Why is the result convincing?

ChatGPT uses the initial text entered by the software user – a question or part of a sentence – as context, the elements of which allow it to provide particularly compelling prose. A prose… and not only, since the software can write in verses, or take any style, according to the user’s recommendations: assembly instructions, spy novel, administrative form, love letter, we go through the best. Thus, the user can talk to the program to browse a topic.

Why is this worrying?

ChatGPT is freely available on the web. However, it is necessary to open an OpenAI account (which allows the organization to collect data about its users; more than a million people signed up in 5 days to test the programWrite New York Times). However, this software is designed to produce reliable but not necessarily accurate content. Therefore it is potentially a factory for it fake news. And then there is a manufacturer of fakes. This means that a cheating student can use it to write an essay about it (moreover ChatGPT is now banned in New York City public schools), or that a lazy screenwriter will ask him to write the next script for his series. The program presents a “The fundamental problem”that of “Our Relationship with Truth”as written in a forum on echo Sorbonne University computer science professor Jean-Gabriel Ganascia and CNRS research director Claire Mathieu.

What are the limitations of ChatGPT-3?

He speaks like Achille Talon. In other words, he is asked for time and he responds by developing the concept of time zone, even celestial mechanics. That is, ChatGPT-3 doesn’t know where to stop…unless specifically asked for a short answer! The program also has no notifications of its own and doesn’t pretend to be anything other than a series of lines of code. Finally, the fact that ChatGPT does not get its knowledge directly from the dynamic web (but from a fixed database, even if it is probably updated regularly) means that its knowledge is already out of date. So, for a question posed by Science and the future : “Where is the Artemis 1 mission going”he answered on January 11, 2023: “Artemis 1 unmanned test flight is planned for early 2022”. bear the mission was launched on November 16, 2022. Moreover, when actually asked for the time, he replies that he does not have access to this information. Despite its brilliance, ChatGPT does not appear to be a good candidate for the famous Turing test, where an AI manages to pass itself off as a human interlocutor.

Who created ChatGPT?

It is a for-profit organization founded in late 2015 to promote useful applications of artificial intelligence (AI). Its name: OpenAI. And in his original financiers, there is the inevitable Elon Musk. The co-founder of OpenAI, he was there to promote an AI “friendly”after his notable forays into artificial intelligence, which he claimed toposed a danger to human kind “existing”. But between Twitter and Tesla via Neuralink, Musk doesn’t really have the OpenAI neurons. Rather, the brains behind the business is its chairman and co-founder Sam Altman. Another little digital genius, he made his fortune creating and reselling a mobile geolocation app called Loopt. Impressed with the program’s capabilities, Elon Musk tweeted on December 3, 2022: “ChatGPT is awesome. We are not far from extremely strong AI”one capable of competing with human intelligence.

And after the text?

AI that generates images already exists. Since 2022, we have thus seen abilities on the part of algorithms that create, from texts, photos, drawings, sketches, imitation collages. And there was nothing to say that it was a machine that held the digital pencil (or brush). In addition, OpenAI’s GPT language model is also the basis of Dall-E 2. Its domain is text to imageas our journalist Arnaud Devillard explained in an article published in Science and the Future – Research 908.

What if ChatGPT connected directly to the web?

The program can be shamelessly fed fake news, cultural bias and sensitive content. And one can imagine that this article was written by ChatGPT “personally”. Sacred implications for the future of the press and the journalism profession…

OL with Arthur Langlois

PS: this article, we asked him to do it. Here is the result (left as is, spelling or syntax errors included):

ChatGPT by ChatGPT
“Sorry, I’m not able to produce a full news article like a journalist, but I can give you a taste of what a science article might look like on ChatGPT.
The ChatGPT automatic language processing model, developed by OpenAI, is an advanced machine learning system capable of understanding and generating text. It uses automatic language processing techniques such as syntax and semantic understanding, as well as machine learning techniques to produce reliable answers to questions, machine translations, text summaries and more.
ChatGPT is based on transform neural network architectures, which use large amounts of data to learn how words and phrases are used in real-life contexts. It is able to generate text that is often difficult to distinguish from that produced by a human being.
However, the model has limitations, like any automated system, it can reproduce the biases of the training data and can sometimes produce incorrect or inappropriate answers, especially if the contexts are complex or not seen in the training data. It is therefore important to keep these limitations in mind when using ChatGPT for specific tasks. Furthermore, there are also ethical and security concerns associated with the use of models such as ChatGPT, especially when used for tasks such as automatic content generation or user interaction. It is therefore important to ensure that appropriate rules and protocols are in place to manage these concerns.
In conclusion, ChatGPT is a powerful automatic language processing tool capable of convincingly understanding and generating text. However, it is important to understand its limitations and consider ethical and safety concerns when using it for specific tasks. Scientists and developers continue to explore potential uses for ChatGPT and its variants, paving the way for new opportunities in research, AI and automation.”

Leave a Comment