¿Adiós a la IA? Los expertos señalan que se queda sin datos para entrenar
Artificial intelligence relies on vast data to identify patterns and learn effectively. While organizations can train models using internal records, they increasingly depend on external internet sources. This reliance has become critical as the industry grapples with the growing scarcity of high-quality human-generated information needed for robust algorithmic development. The central conclusion is that the internet has reached its limit regarding available data for training. Key leaders in the field warn that we have essentially exhausted the cumulative sum of human knowledge available online. This bottleneck threatens the continued advancement of AI capabilities, as the previous strategy of scaling up data ingestion is no longer sustainable due to physical limitations in accessible content. This situation is highly relevant to open data, highlighting a shift toward synthetic datasets to bypass scarcity. However, while synthetic data is cheaper and generated by the AI itself, it creates closed loops that can amplify errors. The transition away from open, diverse human data underscores the urgent need for high-quality, freely accessible datasets to maintain model accuracy and prevent systemic degradation in future AI systems.
Source: larazon.esPublished on 2025-01-11
Related news
- Running Out of AI Training Data? Elon Musk Claims The World Is Facing a Shortage at CES 2025
- Why Synthetic Data Will Transform AI Data Licensing
- Study on medical data finds AI models can easily spread misinformation, even with minimal false input
- IDnow to create synthetic data for EU-funded AI ethics project | Biometric Update
- Nvidia unveils Cosmos, a platform for accelerating the development of AI models in the physical world – NaturalNews.com