For some time now, the Internet has taken over a new tool: ChatGPT, an artificial intelligence in the form of a chatbot. Here’s everything you need to know about it.
On Twitter, Twitch, Reddit Where Youtubemany influencers, content creators and people proudly display their discussions, sometimes completely crazy, with a rather special “chatbot”: ChatGPT. Basically a scientific tool, it is accessible to the general public from this address: openai.com/blog/chatgpt/. A trained language model, ChatGPT is able to generate answers to just about any question you ask yourself. But how does it work? How to use it ? And above all, can we trust his answers? So many questions that we will answer below. Here is everything you need to know about Open AI’s chatbot: ChatGPT, which recently received a $10 billion offer from Google. ChatGPT or the AI that is shaking Silicon Valley.
On November 30, 2022, Open AI, a company specializing in the research and deployment ofartificial intelligence (IA) makes available to the public on the Internet, a somewhat special tool: ChatGPT. If we use the terms of the company, it is then, roughly, “a conversational tool that answers questions, admits mistakes, challenges incorrect premises, and rejects inappropriate requests“. In short, it is a “chatbot” that responds to your requests.
Is ChatGPT really a chatbot?
However, it’s not a “chatbot” like you can find on some e-commerce sites who will help you make your purchases by suggesting certain products or who will direct you to an after-sales service procedure, for example. In effect, their artificial intelligence is only a suggestion toola bit like when the “youtube bot” takes care of showing you the videos you might like on its home page or in the sidebar.
But then what differentiates ChatGPT from the chatbot of Darty or Spotify? Well technically it’s not a chatbot. In effect, Open AI’s tool is a text prediction model that “has been adapted to generate chatbot responses“. It is therefore an AI model which, through its generation of text, is capable of resembling a chatbot, without actually being one. It will therefore generate textual responses word by word using its prediction system. A little what we find on the keyboard of our smartphones, when our device kindly offers us a few words that could be the logical continuation of our sentence.
As proof, here is how AI simply describes itself. We spare you the ultra-scientific details.
What you have to remember is that ChatGPT is a tool created by humans to interact with humans according to what he knows and what he is given to learn.
What is ChatGPT capable of?
In general it can be said that ChatGPT is able to generate anything. It’s a bit vague, we grant you. So here is a non-exhaustive list of the possibilities offered by the Open AI tool:
- Essay plan
- Fictional scenario
- Summary of facts
- Dialog generation
- Alternative Reality Scenario Invention
- song parody
- Code generation
And that, he does not do it only in French. Indeed, Open AI being an American company, the “native” language of ChatGPT is English. However, being a scientific tool, supposed to collect human feedback from all sidesthe response generator is able to understand, interpret and respond in Spanish, Italian, Portuguese, Chinese, Japanese and German.
For the joke, we asked him if he could guide us on the best build to start on Elden Ring, GOTY 2022. We are not experts on this game, but his answer has something to be convincing. We will put his advice into practice to see if he is telling the truth.
Alright, now we know what ChatGPT is, broadly. Let’s find out how you can use it.
How to use ChatGPT?
To use the Open AI tool, it’s very simple. Just follow this short tutorial:
- Go to this address: openai.com/blog/chatgpt/
- Click on “Try ChatGPT”
- A window opens and asks you to register
- So inform Your email address, prove that you’re not a robot (it’s kinda funny when you think about it) then create your password
- Validate the operation
- You then arrive on the interface below (we have put it in dark fashion but you can choose to keep it in normal mode)
- To start a first conversation with ChatGPT, type a question (whatever you want as long as it respects the law and intellectual property) in the text bar provided for this purpose at the bottom of the page
- To create a new conversation, you can click on “New chat” at the top left of your screen
That’s it, you have the cards in hand to use ChatGPT, know that if he does not have the answer to the question or if he cannot accede to your request, he will apologize. You can still try to force him to do so by asking him insistently, but since he obeys the instructions programmed in its code, some things may not be possible. Finally, remember that its learning stopped recently, in fact, not being connected to the Internet, ChatGPT cannot update itself or fetch information from the web. As stated on the home page, it will surely have “limited knowledge about the world and events after 2021“. Now there is one last question to ask.
Can ChatGPT answers be trusted?
Inevitably, when we talk about artificial intelligence, many questions are raised.
- Is she sure?
- How was she trained and by whom?
- Does it respect the laws of robotics?
A few questions that allow you to know where you are stepping and especially if you can use an artificial intelligence without risk. One thinks in particular of the various innovations coming straight from the head ofElon Muskconcerning Tesla for example and its desire to make autonomous cars capable of reacting to their environment. Here, with ChatGPT we do not risk ending up in a tree or a wall at 130 km/h in our Tesla since it is just a text generator in the form of a chatbot. However, can we trust the information that ChatGPT gives us?
The answer is quite simple: Yes and No. Indeed, even though it was trained by the best developers in the world, it is just a robot that can line up words based on previous words. Thus, if by mistake, he gave you erroneous information, it is possible that the rest of your text is too.
Either way, keep one thing in mind: whether it’s what you read on the internet, ChatGPT answers, or answers given by a human, do not take what you are given at face value and do your checks.
Moreover, even Open AI took the lead in informing users, right from ChatGPT’s home page that the tool could “generate incorrect information“. So if you use it to write your dissertations, an article or just for your general knowledge, remember to check the information given to you. Do not hesitate to contact the Open AI teams to point out the errors you have noticed, this will improve the tool so that everyone can take full advantage of its capabilities.
Read also :