002230

GOOGL

80020

00020

GOOG

<p>Oriol Vinyals, co-head of Google Gemini, discussed the future of AI models in a DeepMind podcast, emphasizing algorithm generality and data limitations. He pointed out that AI training includes two stages: pre-training and reinforcement learning. Although the details differ, the principles remain unchanged. He believes that the key to advancing AGI lies in giving computers a digital body to achieve more complex tasks and personalized services. At the same time, he mentioned that the accumulation of innovation is crucial for model iteration</p>

- Oriol Vinyals, VP at Drastic Research, discussed AI model advancements in a Google DeepMind podcast, emphasizing the importance of algorithm generality and the limitations of data in pre-training.  
- He outlined the two fundamental training steps: pre-training and reinforcement learning, noting that while the field has evolved, core principles remain unchanged.  
- Vinyals highlighted the need for innovative approaches to push the boundaries of AI capabilities, particularly in integrating diverse data sources for improved reasoning and personalization.

Wallstreetcn

IFLYTEK

Alphabet

SENSETIME-WR

SENSETIME-W

Alphabet - C

Enlight Media

Hershey

Sunfar

BONA

Heyuan Gas

PERIC

FLAC

DAMAI ENT

Sunwave

GGEC

Hbmc

Grandit

Google Gemini head explains: How to understand the next step of AI?