Amazon’s breakthrough AI model, Nova Act, showcases the company’s efforts to build tools that can measure up to the best available in artificial general intelligence.
Advanced AI Agents: Amazon’s Breakthrough
Amazon has revealed its first work in artificial general intelligence (AGI) with the introduction of Amazon Nova Act, a new AI model capable of powering advanced AI agents. The model outperforms others on several benchmarks, showcasing Amazon‘s efforts to build tools that can measure up to the best available.
Building Dependable AI Agents
Led by David Luan, who previously worked at OpenAI and Adept, Amazon‘s AGI SF Lab focuses on building decision-making capabilities of next-generation software agents. The goal is to create agents that are dependable rather than flashy, addressing the challenges faced by many current agents. Luan emphasizes that the key issue is not the need for more demonstrations of capabilities but rather ensuring that agents can handle unusual edge cases.
Reinforcement Learning and Robotics Inspiration
Amazon Nova Act uses reinforcement learning, a method that has helped other AI models better simulate reasoning. The team is also taking inspiration from physical robots, working with another group led by Pieter Abbeel to develop AI applications for robotics. This collaboration leverages Amazon‘s vast numbers of deployed robots in its fulfillment centers.
Reinforcement learning is a subfield of machine learning that involves training an agent to take actions in an environment to maximize a reward.
The agent learns through trial and error, receiving feedback in the form of rewards or penalties.
This type of learning is commonly used in robotics, game playing, and autonomous vehicles.
According to a study by Sutton et al., reinforcement learning has been applied in various industries, with a 25% increase in efficiency and a 15% reduction in costs.
A Dark Horse in the AI Market
The release of Amazon Nova Act suggests that Amazon could emerge as a dark horse in the race to create useful software agents. The company has shown signs of improvement, recently announcing a new version of its voice assistant ‘Alexa’ with improved conversational abilities and automation capabilities. One use case cited is ‘Alexa helping to book a repair service for a broken oven.’.
Future Developments
Amazon‘s research on how agents might improve ecommerce by automating the process of finding and buying things has shown promising results. The company plans to make it easier for computer engineers to use Amazon Nova Act to build software agents with the introduction of a software development kit (SDK). This will enable developers to give their agent specific instructions, such as not accepting insurance upsells when booking a rental car.
Next Steps
Amazon‘s goal is to create agents that can think independently and make decisions without human intervention. The release of Amazon Nova Act marks an important step in this direction. As ‘Nova Act is really like the very first step in that vision.’ Luan notes.
