On Friday, July 3, we partnered with a few dozen international schools and students from around the world to start closed alpha testing of our systems in the wild. Slanter was built with the objective of bringing AI to schools while prioritizing ethics, truthfulness, and reliability, qualities that current AI systems too often lack.
Slanter is a next-generation AI assistant, built on our research into more targeted training algorithms, optimized tool use, and ethical and honest AI systems. Slanter is accessible through both a chat interface and through our institution platform, and is capable of a variety of conversational and multimodal tasks while maintaining a high degree of accuracy and predictability.
Slanter can help with a multitude of use cases, including research, visual explanations, large-document retrieval, collaborative tasks, curriculum-specific Q&A, coding, and more, through its wide range of tools and exceptional reasoning abilities. Early users report that Slanter produces more relevant answers to curriculum-specific questions and is more open about tasks that raise moral or ethical concerns.
Thanks to more efficient inference architectures, Slanter is both faster and significantly cheaper. These improvements allow Slanter to be accessible to more people than ever, enabling widespread adoption across many educational fields. As we develop these systems, we'll continuously make them more aligned, honest, and helpful, as we learn from our research along the way.
We are excited to see the applications of Slanter across education, research, and many other fields, as we work on a public release.