Scale AI leads groundbreaking research to build safer, more capable AI systems through innovative approaches in post-training optimization, agent development, and evaluation frameworks. Their comprehensive work spans from improving model performance and reliability to developing robust safety measures, all while maintaining a commitment to open collaboration and industry-wide advancement. Through the Safety, Evaluations, and Alignment Lab (SEAL) and various research initiatives, Scale AI is shaping the future of responsible AI development.

Scale’s AISI-approved AI model evaluations are setting a new standard for pre-deployment testing. By offering voluntary, efficient, and third-party validated assessments, we are empowering AI developers to create more reliable models—without the complexities that typically slow down the process.

Text2SQL systems promise to democratize access to enterprise data but often fail to handle the complexity of real-world database queries, even if they perform well on test datasets. We found that Reinforcement Learning from Human Feedback (RLHF) is a viable approach for active learning from incorrect production queries to improve Text2SQL accuracy.

Scale AI and the Center for AI Safety (CAIS) are proud to publish the results of Humanity’s Last Exam, a groundbreaking new AI benchmark that was designed to test the limits of AI knowledge at the frontiers of human expertise.

On January 21st, 2025, Scale AI Founder and CEO, Alexandr Wang sent a letter to President Donald J. Trump on what America needs to do to win the AI war. Read the blog to see the full letter.

As we return to work after the holiday break, the Scale AI Public Sector team wanted to reflect on our work heading into 2025. As strategic rivalries continue to intensify and adversaries form new alliances globally to challenge U.S. leadership in AI, the mission of Scale’s Public Sector team has never been more vital. We are dedicated to ensuring that the U.S. and its allies have the best technology to lead in this increasingly complex global landscape. The snapshot below captures a few key highlights from last year:

Today, TIME is introducing a generative AI experience for their Person of the Year, as well as Person of the Year for 2021, 2022 and 2023. TIME AI will redefine how readers engage with journalism, offering an entirely new way to experience TIME’s vast archive and dynamic storytelling.

Since launching in 2023, Outlier has grown quickly and now provides tens of thousands of contributors the opportunity to earn supplemental income as they help to shape the future of groundbreaking AI technology.

The Scale AI team investigates various synthetic data generation strategies for fine-tuning LLMs under different constraints, providing a robust framework for enterprises to evaluate the most optimal and cost-effective data strategies for fine-tuning LLMs.

Through Outlier, a part of the Scale family of products and services, we connect experts from across domains and around the world with opportunities to train and evaluate generative AI