Amazon Unveils Frontier AI Models and a Customization Tool for Customers
Amazon has unveiled a new family of frontier artificial intelligence models, including two improved large language models, Nova Lite and Nova Pro, a real-time voice model called Nova Sonic, and an experimental model called Nova Omni. These models are designed to be highly customizable, allowing customers to build their own specialized frontier models using a tool called Nova Forge.
The Nova models are the second generation of Amazon's AI offerings and are being made available to a limited number of customers today. While they are not as popular as those offered by rivals like OpenAI and Google, Amazon's plan to make them highly customizable could see them gain traction with its cloud users.
Nova Forge is a significant innovation, as it allows customers to add their own training data to unfinished versions of the Nova 2 Lite and Pro models at various stages of model training, including the process of building the base model. This approach empowers customers to create specialized models tailored to their specific needs, a capability that is typically reserved for large AI labs.
One customer that has tested Nova Forge is Reddit, which used it to create a custom model to identify content that breaks the platform's rules. The model, built through custom pre-training and conventional fine-tuning, is an expert at understanding and using Reddit, according to Reddit's chief technology officer, Chris Slowe.
Nova Forge offers a new approach to AI customization, although it is locked into Amazon's cloud. Building a large language model from scratch can be costly, but Nova Forge is expected to be significantly cheaper. Amazon is also investing billions in AI infrastructure, positioning itself as a dark horse in the AI race, with a portfolio of advanced AI capabilities and generative AI integration into its shopping platform.
The customizable nature of Nova models is highlighted as a key feature, with potential applications in content moderation and other specialized tasks. As companies seek tools beyond the capabilities of general-purpose models, Amazon's approach may prove savvy, especially with the growing demand for AI and the challenges of building custom models.
