Back
Vishu Bheda
ย โขย
Medialย โขย 5m
Jensen Huang, the CEO of NVIDIA, describes how AI is advancing in three key dimensions: 1. Pre-training: This is like getting a college degree. AI models are trained on massive datasets to develop broad, general knowledge about the world. 2. Post-training: Similar to specializing in a career, AI is fine-tuned to become an expert in specific fields like healthcare, finance, or language. 3. Test-time compute: This represents "thinking on the spot." When AI is deployed, it uses its knowledge to solve real-world problems in real time, adapting to new challenges. These three stagesโlearning broadly, specializing deeply, and thinking quicklyโare driving the rapid progress of AI. Huang highlights that this multi-dimensional development is why AI scaling continues, enabling smarter, faster, and more efficient systems capable of solving increasingly complex problems in every field.
Replies (5)
More like this
Recommendations from Medial
Download the medial app to read full posts, comements and news.