Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Elon Musk agrees with other AI experts that there is too little real data to train AI models.
“Now we’re tired of a lot of people’s knowledge….in AI education,” Musk said in an interview with Stagwell chairman Mark Penn on X late Wednesday. “This happened last year.”
Musk, who owns the AI company xAI, also addressed the previous topics of OpenAI chief scientist Ilya Sutskever. to touch at NeurIPS, a machine learning conference, in an address in December. Sutskever, who said the AI industry has arrived at what he called “super data,” predicted that the lack of training data will force them to abandon the methods used today.
Indeed, Musk has pointed out that artificial data — that which is generated by AI models itself — is the way forward. “With generated data … (AI) will automatically create itself and go through this self-learning process with generated data,” he said.
Other companies, including tech giants such as Microsoft, Meta, OpenAI, and Anthropic, are already using synthetic data to train AI models. Gartner comparison 60% of spending on AI and analytics projects in 2024 will be artificially generated.
Microsoft is Phi-4which opened earlier on Wednesday, was taught on artificial elements along with virtual reality. The same goes for Google Gemma examples. Anthropic used artificial intelligence to develop one of its most successful systems, Claude 3.5 Sonnet. And Meta changed his latest version Llama list of examples using AI generated data.
Training on artificial intelligence has other advantages, such as cost savings. The first AI author says that his version of Palmyra X 004, which was created using almost all artificial sources, only cost $700,000 to make – in comparison compared to $4.6 million for similar OpenAI models.
But there are also disadvantages. Another study he points out that artificial data can lead to the collapse of a model, where the model is less “artificial” – and biased – in its results, ultimately undermining its performance.