Elon Musk, who has been fighting artificial intelligence and its dangers for a long time, would already be working on one. Sources close to Twitter mentioned that the tycoon bought thousands of graphics cards in order to develop an extensive language model (LLM) within the company. With this, Musk would launch his plan to create an AI that competes with ChatGPT, although with less strict guidelines to avoid censorship.
The sources confided to Business Insider that Elon Musk purchased 10,000 graphics processing units to power his new model. The project of Musk’s generative AI is in a preliminary stage, so there are not many details. Purchasing these teams would be proof that you are committed to carrying out your idea of competing with ChatGPT.
The rumors about a new lab to rival OpenAI They have been dormant for weeks. A report of The Informationno pointed out that Elon Musk would have recruited Igor Babuschkin, a research engineer who has worked on DeepMind and OpenAI. Although Babuschkin loves the idea of working for Elon on a new LLM model, the engineer stated that he has not joined the project, since there is no concrete development plan.
The truth is that a generative AI developed by Elon Musk does not sound far-fetched if we consider the mogul’s frequent attacks on ChatGPT. Musk declared in February that OpenAI was created as a non-profit company to serve as a counterbalance to Google. “It has now become a closed, top-profit company effectively controlled by Microsoft. Not what I intended at all,” he said in a tweet.
Elon Musk’s ChatGPT would kill off AI woke
Elon Musk too has charged against ChatGPT’s security guidelines. OpenAI’s artificial intelligence has removed sexist responses and adjusted its guidelines to avoid hateful content. That has not gone down well with a sector of users, who accuse the company of turning the GPT model into woke and ask their creators to consider an alternative.
In a tweet posted in December 2022, Musk expressed that the danger of training an AI to be woke It is mortal.
The idea of a ChatGPT with more relaxed security guidelines sounds like a project that Elon Musk would carry out. However, this idea is not shared by Igor Babuschkin, who defended the tycoon and assured that he has no plans to create an AI antiwoke. Whatever the objective, the truth is that the development of a new model of artificial intelligence would be underway and Twitter is the key.
Elon Musk has at hand the data of millions of users with which he could train his extensive language model. The Twitter database would be a gold mine, since it has millions of messages and the context that gives rise to these expressions. The problem is that this data set carries with it the toxicity and hate speech that distinguishes the social network.
Although Musk has not commented on the matter, it is almost a fact that he plans to develop a generative AI. For now, the tycoon has focused his attacks on OpenAI and wants models like GPT-4 to pause their development for six months to avoid “catastrophic effects for society.” The owner of Twitter and other experts are seeking to create new security protocols to ensure that artificial intelligence systems are safe.
Could it be that Elon cares about humanity or just wants OpenAI to take a break to join the competition?