Human data pool for AI training exhausted in 2024 – Elon Musk

Tesla CEO Elon Musk has stated that all available human data for Artificial intelligence (AI) training, including books, was exhausted in 2024.

This assertion aligns with conclusions reached by other experts in the field.

During a livestream conversation with Mark Penn, chairman of Stagwell, Musk made these comments, which were streamed on X.

According to him, the next viable option for AI training is synthetic data, which is data generated by AI itself.

“AI is advancing in terms of hardware and software, and it is now moving towards synthetic data because we have exhausted all human-generated data. We have literally run out of the entire internet, all books ever written, and all interesting videos.

ALSO READ: Elon Musk changes his name to Kekius Maximus on X

“We have now reached the limit of cumulative human knowledge available for AI training, and this happened last year. The only way to supplement that now is with synthetic data, which AI creates itself. It can write essays, develop theses, and then grade its own work, engaging in a process of self-learning with synthetic data,” Musk explained.

However, Musk acknowledged that using synthetic data for AI training poses its own challenges, particularly in determining the accuracy of its generated answers.

“This is always challenging because it’s difficult to know if an answer is a hallucination or if it’s real. Finding the ground truth is complex,” he commented.

Furthermore, some researchers have warned that relying on synthetic data could lead to model collapse, where an AI model becomes less creative and more biased in its outputs, potentially compromising its functionality.

READ MORE FROM: NIGERIAN TRIBUNE