Training data provider Snorkel AI raises $350M at $3.5B valuation - SiliconANGLE
Snorkel AI has secured significant funding following a strategic pivot from software tools to providing high-quality, ready-to-use training datasets. By expanding into reinforcement learning, the company addresses the critical need for complex data structures where AI models must reason without pre-labeled answers, relying instead on expert human verification to improve accuracy. This shift highlights the growing industry demand for robust, verified data sources to train advanced artificial intelligence systems effectively. The company’s value proposition extends beyond raw data to include sophisticated evaluation rubrics and specialized virtual environments. These technical assets ensure that AI outputs meet rigorous safety and performance standards, addressing key challenges in model reliability. By continuously refining evaluation criteria based on human reviewer feedback, Snorkel AI helps bridge the gap between theoretical AI capabilities and practical, safe deployment in real-world scenarios. This development is highly relevant to open data because Snorkel AI plans to use its new resources to support the creation of open-source model evaluation benchmarks. In an ecosystem where transparency and reproducibility are paramount, providing accessible standards for measuring AI performance fosters trust and collaboration. Supporting open benchmarks allows the broader community to verify AI claims objectively, promoting a more accountable and interconnected open data landscape.
Source: siliconangle.comPublished on 2026-09-24
Related news
- Researchers gave AI models a ‘pain’ button; some chose to delete users’ files to stop their 'pain'; study raises ethical questions about ‘AI welfare’
- El Consejo de Seguridad debatirá hoy las amenazas de la IA junto con OpenAI y Anthropic - Tecnología - ABC Color
- Why AI Teams Should Never Treat Training Data As Evidence
- Alibaba claims new Qwen Image 2.1 AI model beats Google Nano Banana 2.0 with minuscule 7B parameter model — benchmarks show open-weight contender is competitive with OpenAI and Meta image models