Nvidia's senior director of generative AI software, Joey Conway, discussed the growing potential of local and open AI models. These models are becoming increasingly capable and can run on desktop computers. This shift is changing how organizations approach AI.
The models small enough to run on a desktop are improving rapidly. Conway explained that the focus is now on what can be done with these models and how to maximize their utility. Nvidia is exploring ways to make the most of both local and frontier AI models.
Conway highlighted that local models offer advantages in terms of data privacy and reduced latency. They can be used for tasks that require immediate processing and can be tailored to specific organizational needs. Frontier models, on the other hand, are more powerful and can handle complex tasks. Nvidia is working on integrating both types of models to provide a more comprehensive AI solution.
Nvidia believes that a hybrid approach is the future of AI. „We love the world where we can use both,”Conway said, emphasizing the benefits of combining local and frontier models. This approach allows organizations to leverage the strengths of each type of model.
While local models are becoming more capable, they still have limitations. Conway noted that certain tasks require the power of frontier models. However, local models can be used for a wide range of applications, from data analysis to content generation. As local models continue to improve, they are likely to become an increasingly important part of an organization's AI strategy.
The growing capability of local AI models will have significant consequences for organizations. As these models become more powerful, they will enable businesses to make more efficient use of their data and improve their overall AI capabilities.