20VC: Why Foundation Model Performance is Not Diminishing But Models Are Commoditising, Why Nvidia Will Enter the Model Space and Models Will Enter the Chip Space & The Right Business Model for AI Software with David Luan, Co-Founder @ Adept
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
Adept’s CEO on Memory in LLM and How It Affects Product Development
Memory in artificial intelligence can be divided into short-term working memory and long-term memory. Progress has been made in short-term working memory, with models like Gemini having a context length of a million tokens, enabling tasks like generating step-by-step instructions from video snippets. However, the challenge lies in long-term memory, where the focus should be on application developers incorporating user preferences and experiences into the software system using large language models (LLMs). It is emphasized that LLMs themselves are not the end product, but rather a component of the larger software system. Application developers should be empowered to embed long-term memory about user preferences, such as personal flight seat preferences, within their systems, making the entire software more intelligent and user-centric.
Remember Everything You Learn from Podcasts
