For their virality on social networks, “fake news” has become the haunt of Democrats. This phenomenon is so dangerous that bills appear against these “fake news” capable of manipulating the masses.
But this latest version of AI (artificial intelligence) developed by OpenAI,, a nonprofit artificial Intelligence research company backed by Elon Musk is a text generator that has ranked as the head of an information aggregator. But it’s not just texts, no ! But downright news reports, Press articles and even fiction.
So far everything is fine, except that imagine that the machine is able to create “deepfakes for next” in other words false information more than credible created by an artificial Intelligence and even share them himself on social networks.
American media specialists have tested this famous AI that everyone is talking about. This is the case in particular of wired magazine who had fun to write an article by typing just at the beginning ” Hillary Clinton and Georges Soros “’ we do not ignore that many fake news circulated about the former Democratic candidate and the famous American financier).
So Artificial Intelligence invented a whole story of coalition between the two using an audio recording, indeed a beautiful lie!
Like any artificial Intelligence, such a program feeds on existing elements. We can thus deduce that it was based on false information found here and there or poorly understood information. In truth, an artificial Intelligence understands nothing of everything it tells. This is the reason why she invents stories ‘yogurt pancakes “or fire under the sea”.
GPT-3 : what is the AI developed by OpenAI ?
GPT-3 : l’énorme IA qui alarme !
Artificial Intelligence GPT – 3 became known at lightning speed via its viral posts on social networks. Between hype and anxiety what exactly is GPT-3 ? What is the reason why this Artificial Intelligence surprised so many developers in particular ? Does this change anything for the future ?
For science yes !
The company behind this technology is OpenAI, and it is famous for three main reasons.
- The first is that one of the founders is none other than the famous Elon Musk, the person who is most afraid of Artificial Intelligence in the world, the other founder is Sam Altman.
- The second reason is the reason for the existence of this firm. The goal of OpenAI is to develop AI in a certain and sustainable way for humans. Musk is convinced that uncontrolled artificial Intelligence will cause the end of humanity, so OpenAI serves to promote a safe way to create it.
- The last reason OpenAI frequently does talk about it with its remarkable repeated successes in its field. Notably the older GPT and GPT2 models.
Generative Pretrained Transformer 3 for short GPT3 is a language model using deep learning (neural network) to produce text.
The principle of a language model is to look at a part of a sentence and produce the next word.
GPT – 3 is a technology that has made ‘swallow’ a huge corpus of text, the largest ever recorded. Around a thousand billion words via Common crawl.
Based on this tiny sample, it generates text (output) that makes the most sense for a request (input). Just by learning to predict which letters and words tend to follow one another when a human writes them, GPT – 3 is able to predict a text that can make sense in a context.
It’s a bit like your mobile that gets to predict the next word when you write a message. By contrast, it is infinitely bigger, more complicated and more evolved.
But so far GPT – 3 does not bring something new. The architecture of this technology (transformer-based) is the same base as the previous version (GPT-2) since this architecture is almost 3 years old.
You’re probably wondering what’S so special about GPT-3.
GPT-3 : Beyond mesure!
Well yes ! GPT – 3 is absolutely gigantic ! GPT – 2 is already out of standard with no less than 1.5 billion parameters. The previous record being held by Turing-NLG Microsoft with its 17 billion parameters.
But hold on, GPT – 3 is coming in to blow up all records with a “pretentious” figure of 175 billion parameters. Never seen or heard !
What types of texts is AI GPT-3 capable of writing ?
Who speaks of outsized size will surely not fail to talk about outsized results. GPT – 3 is capable of far too many things: indeed it can have different roles: between translator, poet, author, programmer it is able to generate different types of texts namely :
- works of fiction, essays or poetry ;
- scientific articles ;
- medical diagnostics ;
- whole fictions ;
- cooking recipes ;
- song lyrics ;
- emails with your own style ;
- blog posts about himself ;
- number one blog posts about a hacker news ;
- tweet ;
- a chatbot involving historical characters ;
- computer manuals ;
- lines of code in HTML or other programming languages…,
In short the list is too long and keeps getting longer as the days go by.
And GPT – 3 is also capable of something else that immediately worried the developers. GPT – 3 is capable of producing code.
Will you like a todo list app in ReactJS instantly ? Just ask.
All this to say that some small easy tasks that developers do can be automated by GPT-3, wonderful right ?
One problem :
Is the commercialization of AI safe ?
As we speak, OpenAI refuses to give the source code despite its goal of transparency.
The official reason ?
This technology is too big and “greedy” for people to run it themselves. They put it behind a paywall that allows OpenAI to monetize its research.
Two months after the first appearance of GPT-3, many people agree on one thing : GPT-3 is as superb as it is overestimated. This is especially true for developerss.
GPT cannot be denied being super impressive because it remains a great tool to do many things. But very unfortunately the important thing to understand is that GPT – 3 does not understand anything about the content it creates. This technology makes a mathematical prediction of how certain letters should appear in a certain order.
It should be noted that the work of a developer is infinitely very complicated. Thus developing a complete application that corresponds to a human need requires a lot of thought.
No algorithm has the ability to compete with Man if it comes to thinking about problems. Taking our thinking further, we realize that GPT – 3 is simply a new tool for developers.
It’s almost a new language where the interface would be directly explanations of what needs to be done in place of the code. And it is quite relevant to compare the instructions given to GPT – 3 with code. GPT – 3 does not understand anything to vague requirements.
What GPT – 3 needs are very precise, surgical instructions, otherwise it risks producing anything.
Where a human being will translate, correct and arrange vague requests, GPT – 3 will produce ‘shit’. See even do nothing at all.
GPT – 3 is the largest search table ever created. Super intelligence is another story. That’s all but for today and Sam Altman, the second Founder of OpenAI, admits it himself.For now, the future of programming remains programming.
Artificial Intelligence continues to make its way, but until it understands what it is doing, it is not likely to do so many wonders. In the end, you just have to admit to her that she is capable of more and more mind-boggling, destabilizing things.
But tell us, how can you be sure that this article was not written by an artificial intelligence?
Read similar articles!