What is ChatGPT, the chatbot that impresses the Internet and asks about the progress of AI?

It is the best algorithm of the moment when it comes to simulating human conversation. chatbot ChatGPT never ceases to impress on the internet. Designed by American startup OpenAI, founded in 2015 in the United States with the aim of promoting and developing artificial intelligence with a human face, this chatbot marks a new stage in the advancement of automatic natural language processing algorithms . Launched last week by the OpenAI association, the tool, still in the testing phase, gathered on Tuesday December 6 more than one million users who have already tested the robustness and complexity of its answers.

The latest project of the OpenAI association, to which we already owe the DALL-E model, capable of generating original images from plain text, ChatGPT is a chatbot capable of understanding questions and answering them in natural language with unprecedented quality and in multiple languages, including French. This program relies for this on an artificial intelligence, GPT3, which uses machine learning algorithms fed with millions of texts and conversations available on the Internet (press articles, academics, blogs, movie scripts, TV shows, etc.) to understand meaning of words and phrases and generate appropriate responses based on context.

What can ChatGPT do?

ChatGPT covers every topic imaginable. After there was created an account on the OpenAI website, you can ask him to give you a cooking recipe, write an essay, write a cover letter for a certain company and activity, write a song on a certain topic or invent a story to shown to the children in the evening. ChatGPT can also answer particularly complex requests, to debug computer code, solve a math problem, or write an article on a topic of your choice. Regardless of the question, within the bounds of legality, ChatGPT will provide you with advanced answers in just seconds. And he continues to learn from the test of Internet users.

Chatbot ChatGPT easily provides an essay answering the final subject of the French degree:

A dissertation corresponding to the final subject of the French degree

Credit: OpenAI / RTL

A cooking recipe from leftovers in the fridge:

A recipe for cooking using food from the refrigerator

A recipe for cooking using food from the refrigerator

Credit: OpenAI / RTL

An article introducing ChatGPT and the questions raised by advances in human activity intelligence:

We asked chatbot ChatGPT to write an article about it

We asked chatbot ChatGPT to write an article about it

Credit: OpenAI/RTL

Provoking humanity to question the progress of artificial intelligence

Still in the experimental stage, ChatGPT already raises a number of questions about its current and future capabilities. Some already see it a future competitor of Google, capable of formulating more complex and contextualized responses to Internet users’ questions. Others ask about the threat it poses to human activities based on content creation, for example, such as press officers or journalists. It can also be used for plagiarism purposes by students for schoolwork. It is even diverted to catch people spreading false information or creating believable hoaxes.

However, the tool still has room for improvement. “It’s important to note that ChatGPT is not yet able to fully replace humans in content creation. It still cannot understand the intent behind a text and still needs to be trained on large amounts of data to improve,” replied the robot when questioned. ChatGPT is also unable to gather first-hand information, conduct interviews, or use nuance and critical thinking to justify its choices.. At least in its initial version.

“Future developments in artificial intelligence may be able to make ChatGPT more human in its ability to create content,” says the chatbot created by OpenAI. “This raises ethical questions about the use of AI in the fields of content creation and how this may affect human workers. It is important to continue to monitor these developments and consider the implications for the future.”

In addition to the OpenAI association, American companies have also designed particularly advanced experimental chatbots. The character. him, a company founded by former Google engineers, released a chatbot in October that can take on any person with attributes and behaviors considered authentic by humans. Meta researchers were developed recently a computer program called Ciceroas the Roman statesman Cicero, who has proven himself in Diplomacy, a board game that requires negotiating skills.

The editors recommend

News from the RTL editorial office in your mailbox.

Thanks to your RTL account, subscribe to the RTL newsletter to follow all the news on a daily basis

Leave a Comment