English 
Spanish 
3 min read
5 comments
Billionaire Elon Musk says human data for training AI has run out and we are entering the 'peak data' theory. Find out the best solution proposed by the tycoon.
A few weeks ago, experts pointed out that the world had reached the limit of human data to train AI. In the wake of the popularity of Chat GPT, many companies have tried to imitate its capabilities, with companies like Google, Apple, and Meta working to offer generative AI assistants. In this article, we will understand what billionaire Elon Musk calls the 'peak data' theory.
Billionaire Elon Musk presents solution to 'peak data' theory
The most recognized voice of technology sector revealed that human data to train AI is no longer available and that a solution is urgently needed. According to the website TechCrunch, billionaire Elon Musk stated that we have reached the limit of real-world data to train AI models. However, he emphasizes that this situation did not happen now, but rather a few months ago, still in 2024.
In this way, the tycoon reinforces the words of Ilya Sutskever, former chief scientist at OpenAI, who had already indicated in 2022 that the sector had reached a situation he called the 'peak data' theory. In this way, billionaire Elon Musk presented a temporary solution to further train AI models.
For Musk, a great option to avoid using human data to train AI is to use the data generated by Artificial Intelligence itself, an aspect known as “synthetic data”. Thus, machine learning chains will be created, following in the footsteps of companies such as Microsoft, Meta, OpenAI and Anthropic, which are already on this path.
Challenges when using synthetic data to train AI
In fact, some estimates believe that by 2024, 60% of the data used will be synthetic, since, in addition to everything mentioned above, there will be other advantages, such as cost reduction. The AI startup Writer, for example, developed its model Palmyra X 004 almost entirely on synthetic data, spending just $700 million, compared to the $4,6 million estimated for a similarly sized model from OpenAI.
Still, some research suggests that the end of human data to train AI and the use of synthetic data could lead to the collapse of different models, as creativity would be reduced and biases in the results would increase.
In fact, if the data generated has biases or limitations, the models trained with it will reproduce those same problems in their results. However, this does not seem to be a limitation for either billionaire Elon Musk or companies like Microsoft, Google or Anthropic, as they used it in models like Phi-4, Gemma and Claude 3.5 Sonnet.
Understand the impacts of AI on companies by 2030
Even with the ‘peak data’ theory, by 2030, the concept of “data ubiquity” will be deeply ingrained in the way businesses operate, creating an ecosystem where real-time information not only guides human decisions, but also automates processes with unprecedented efficiency.
The generative AI revolution has brought to light possibilities that were once the province of science fiction. Companies are rethinking their approaches and business models, driven by a relentless pursuit of innovation and competitiveness.
From creating personalized medicines to automated management of complex operations, the opportunities seem limitless. But these promises depend on one essential ingredient: relevant, accessible, and well-structured data. Without this solid foundation, even the most advanced technologies fail to reach their transformative potential.
Musk is a bad character, a **** billionaire. **** man.
And you're the one who's ****, huh. Damn, crazy. You rock. You rock.
Elon, I am with you and registered in the efficiency department and I need a Tesla card and cell phone and car and house and salary, English and Portuguese prince Oscar de Araújo Martins, lawyer, Marshal of America and Venezuela and first Marshal of Brazil. I love you very much, brother, we are children of God.