The truth on the graph announcing a new version of ChatGpt, even more powerful
[ad_1]
ChatGpt, with his prose so close to that of menis no longer enough. Users dream an even more powerful version of artificial intelligence developed by Open AI. That’s not the version for a fee so there is already a waiting list.
Hopefully, rather, the advent of the next large language model (LLM) called Gpt-4. The current ChatGpt is based, we recall, on an ‘improved’ version of Gpt-3i.e. Gpt3.5.
Artificial intelligence
Behind the scenes of ChatGPT and Lensa: what are Laion 5B and Common Crawl and how they work
by Emanuele Capone
To highlight the difference between Gpt-3 and Gpt-4someone also created a chart that instantly went viral. There are two spheres in comparison – one tiny and the other huge – which hypothesize the difference in data they would feed Gpt-3 (175 billion parameters) and Gpt-4 (1 billion parameters). “Gpt-4 will be able to write a 60,000-key book from a single prompt,” wrote one of the Twitter users who posted the graph, without specifying his sources.
Rumors of an imminent arrival of Gpt-4 and its super power have been debunked by Sam Altman, CEO of Open AI, in a recent interview. “These are ridiculous rumors – he said – I don’t know who spreads them. People ask to be disappointed and they will be”. Altman also commented on the viral chart on ChatGpt: “It’s Just Bullshit”.
During the interview given to StrictlyVCAltman said other interesting stuff:
“There will come an AI model that will be able to produce videos. But I don’t want to make predictions. We will try to do it, others are trying too, we may succeed soon”.
“We’re not making much money with OpenAI, we’re still early days.”
Artificial intelligence
“How will the world end?” ChatGpt inventor chases the hardest answer
by Pier Luigi Pisa
“People should be entitled to their version of AI. If you want the model that is never offensive, safe for work, you should get it. And if you want one that is more aggressive, creative and exploratory, but still able to say uncomfortable, or that makes other people uncomfortable, that is creative and exploratory but says some things you may not feel comfortable with, you should have that. I think there will be many systems in the world based on different values. What I really think – but it will take time – is that the user should be able to write a few pages of “here’s what I want, here’s my values, here’s how I want the AI to behave”. act exactly the way it was described, because your AI should be.”
“In the short term we will try to do something that limits plagiarism, something that helps teachers identify work made with ChatGpt. But someone will be able to circumvent that system and therefore I don’t think it is something that the school and society should bet on in future. We live in a new world. AI-generated text is something we should all adapt to”
[ad_2]
Source link
