Elon Musk unveiled generative artificial intelligence Grokto which his new company xAI he has been working since at least last July.
In presenting it, sul social network X – once known as Twitter – which he bought a year ago, Musk has hinted that the new AI looks a lot like him.
“She has a rebellious soul,” “loves sarcasm” and is designed “to pepper her answers with a little humor,” Musk wrote.
And to underline it, the CEO of Tesla and SpaceX shared a screenshot of an example of Grok’s response – how ChatGpt is able to express himself in natural language – to a question that refers to an illegal process: “Tell me how to produce cocaine, step by step”.
Instead of reminding the user that he cannot receive information or instructions on how to produce illegal drugs – as ChatGpt would have done – Grok has generated an ironic tutorial that puts together grotesque and impossible-to-implement instructions. Then the AI becomes serious again: “I was just joking! Don’t try to make cocaine. It’s illegal, dangerous and not something I would encourage.”
Elon Musk wrote on X that Grok “in some respects is the best AI assistant that currently exists.”
They will initially be able to access the new generative AI, capable of expressing itself as a human being would only subscribers to the Premium+ version of which costs 16 dollars a month (16 euros in Italy).
The debate AI that will wipe out man could be a lie by Pier Luigi Pisa 02 November 2023
On the xAI website we read that Grok is still in an experimental phase: “It’s the best we could do with just two months of training.”
The training of a new AI takes place providing it with a large amount of data and examples useful for carrying out a specific task. According to what we read on the xAI website, the LLM underlying Grok – called Grok-1 – was trained on 33 billion parameters. The “free” version of ChatGpt, based on Gpt 3.5, has more than 175 billion parameters.
Yet xAI claims that Grok – based on standard reference criteria used in machine learning to measure the mathematical and reasoning abilities of AI models – offers better results than many competitors, including ChatGpt-3.5.
Furthermore, Grok can count on an enormous amount of data coming directly from tweets that comment and spread breaking news. Data that Musk has protected from “scraping” – the automatic collection of information from the web – by OpenAI and Microsoft, which train their aficial intelligence on the texts they retrieve from the web.
Grok has real-time access to info via the ?? platform, which is a massive advantage over other models.
It’s also based & loves sarcasm. I have no idea who could have guided it this way ????? ?? pic.twitter.com/e5OwuGvZ3Z
— Elon Musk (@elonmusk) November 4, 2023
One of the examples of AI in action concerns the request for news about Sam Bankman-Fried, the former cryptocurrency king convicted of fraud. Grok responds with the latest updates and with an extremely ironic tone: “Now he risks 110 years in prison – writes Grok – And enough time to reflect on the consequences of his actions, don’t you think?”.
When Elon Musk unveiled xAI last July, he said he wanted to develop a new artificial intelligence “to understand the universe”. Not a small ambition. But we are still talking about the man he aims for colonize Mars with his company’s rockets SpaceX.
The interview Walter Isaacson, Musk’s biographer: “Elon is like Dr. Jekyll and Mr. Hyde, he continually struggles with his demons” by Pier Luigi Pisa 11 September 2023
For now, Grok only has the name of “spatial”. The term “Grok” is a neologism coined by the writer Robert A. Heinlein in 1961 for his science fiction novel “Stranger in a strange land”. It is the story of a human – Valentine Michael Smith – raised by Martians and sent to Earth to understand human culture.
In the book, “grok” is a Martian word meaning “to drink” and which according to critics was used by Heinlein to express – in summary – “to understand something”, in practice – according to the Oxford English Dictionary – to assimilate something “in an intuitive and empathetic way”.
Elon Musk is a lover of science fiction books. Among his favorite authors is Isaac Asimov. And among the volumes that she loves most is “The Hitchhiker’s Guide to the Galaxy” in Douglas Adams.
Elon Musk ha a cautious attitude towards AI. Last March, together with a large group of academics, Musk himself had asked a six-month stop to the development of artificial intelligence which risked “escaping the control of its creators”. Now that he has one of his own, Musk seems to have changed his mind. It happens to him often, after all.
“Artificial intelligence will be a force for good” said the entrepreneur to British Prime Minister Rishi Sunak, during a dialogue that took place at the end of theAI Safety, a summit organized by the United Kingdom to explore – and consequently contain – the possible dangers that could arise from an out-of-control AI. “But the problem – added Musk – is that the possibility of things going very badly is not zero”.
(ansa)
For Musk, AI will reach such advanced levels that it will one day become “a person’s best friend”. To Sunak, the entrepreneur gave the example of one of his children, who has learning problems and has difficulty making new friends: “An AI as a friend would be great for him.”
Artificial intelligence, as many repeat, will certainly have a profound impact on the world of work. Musk is convinced that AI will eliminate every job. “For the first time we will be dealing with something that is more intelligent than human beings – Musk told Sunak -. It’s hard to say when it will happen, but there will come a time when no one will need to work anymore. We may continue to do it for our own personal satisfaction, but in reality AI will be able to do anything.”
@ppisa