There is no more data left to train AI

Elon Musk recently announced during a live chat with other AI experts that we have used all available data for AI models and little real data for Education models artificial intelligence is left

According to the TechCrunch report, Musk said during his conversation with Mark Penn, the head of the Stagwell Marketing Company at X:

“Right now we’ve basically used all the gradually accumulated human knowledge we need to train AI. “This technically happened last year.”

Musk, who has his own artificial intelligence company called xAI, echoed a point previously raised by Ilya Sotskur, former chief scientist of OpenAI, at the NeurIPS machine learning conference.

Sotscore said in December that the AI ​​industry has reached what he calls “peak data,” predicting that a lack of training data will change the way models are developed today.

Elon Musk’s proposal to use synthetic data

Musk says that tech companies will continue to have to train models from Synthetic data use, that is, the data produced by the artificial intelligence itself. He says:

“The only way to compensate for the lack of real data is to use synthetic data, where artificial intelligence creates training data. “With artificial data, artificial intelligence somehow grades itself and goes through the process of self-learning.”

Elon Musk mentioned that even now companies like Microsoft, Meta, OpenAI and Entropic are using artificial data to train their artificial intelligence models. The research company Gartner also said in a report that 60% of the data used in 2024 for artificial intelligence and analytics projects will be artificially generated.

But there are also concerns about the use of artificial data. In some researches, it has been shown that their use can cause the collapse of the model, that is, the level of creativity of artificial intelligence models decreases. Also, some experts have said that if the data used to train these models has biases and limitations, this will also affect their output.