Impossible not to talk about ChatGPT at the moment, as the media buzz is strong around OpenAI’s conversation tool. But how does this tool work? What are the mechanisms that work in the “belly of the beast”? What are its potential limitations? And above all the question that dominates today in the landerneau of search engines: can this tool compete with Google in the long term? Sylvain Peyronnet, an eminently recognized specialist in the field, gives here his analysis and forecasts on this subject.
I have been working in algorithms for over 20 years now. My lifelong topic is decision making in the presence of uncertainty and in a context of large amounts of data. Some of the algorithms I worked on are now in what is commonly called artificial intelligence.
Still, I could never have imagined what the past few years have seen in the form of machine learning, natural language processing, image analysis and others. We live in a period where exceptional results are presented almost weekly. To be honest, I feel like I’m reliving the early good years of the web…
The topic of the moment is of course the famous chatGPT. The latest addition to openAI is making a lot of noise and is popular thanks to its ability to interact with humans. No one can deny that the tool is truly amazing, although of course we also see its limits from the first use.
For SEO professionals, the AI revolution is both an opportunity and a risk. It’s a risk for webmasters, but it’s an opportunity for those who have called them and who might see a way to get for less what they’ve paid for elsewhere. For a search engine like Google, this is obviously a risk, since Sundar Pichai, Google boss, has even created a new internal organization in order not to be overtaken in AI by openAI.
In this article I will tell you about chatGPT. What exactly is it? What are its technical limits? How much does it cost ? I will also try to give some answers to the question that bothers us all: is chatGPT a potential danger to Google?
ChatGPT, what is it?
ChatGPT (reference ) is a model of the language that – as the name suggests – is designed to be conversational. This means that he is able to follow instructions given by a human via a fast (a question). To follow these instructions, the model has the ability to engage in dialogue and use a kind of common sense that a human might have, which is not found in other models (like GPT3 for example).
Another strong point of the model, mimicking human behavior, is its ability for memory continuity: chatGPT is able to remember what you told it in previous questions and to elaborate on answers based on this chat history. This aspect is the most anthropomorphic: we sometimes have the illusion of debating with a real person, who will also make mistakes from time to time in his answers.
ChatGPT, this is the latest model in a long line of language models. It all started with word2vec by Tomas Mikolov (then at Google), so quick text from the same researcher (then on Facebook) and many more likes Ernie in 2019 (at Baidu), BERT in 2018 (Google), Grover and Elmo in 2018 and 2019 (Allen Institute). There are also models in France (at Lighton).
Over time, these models have become bigger and bigger and more expressive. But the real breakthrough was the emergence of transformer-based models at openAI. It is now the leading operator in the field, ahead of all others, whether for image or text applications. Their first model dates from 2018, it is the GPT. But the general public started to take an interest in them with GPT2, the first very large model, when it “only” had 1.5 billion parameters and a training data set of 8 million web pages. We now realize that GPT2 was ultimately just a proof of concept and the real breakthrough is GPT3 (see reference ), a model with 175 billion parameters, trained on a data set with several billion pages of content.
[Cet article est disponible sous sa forme complète pour les abonnés du site Réacteur. Pour en savoir plus : https://www.reacteur.com/2023/01/chatgpt-peut-il-detroner-google.html]
An article written by Sylvain Peyronetdesigner of the Babbar backlink analysis tool.