Google DeepMind CEO questions DeepSeek’s AI efficiency claims
Google DeepMind’s CEO, Demis Hassabis, criticizes DeepSeek’s claims about its R1 model being as powerful as OpenAI’s GPT-4 but cheaper to train. Hassabis suggests DeepSeek’s cost figures might not include full development expenses and hints at possible…
- Google DeepMind’s CEO, Demis Hassabis, criticizes DeepSeek’s claims about its R1 model being as powerful as OpenAI’s GPT-4 but cheaper to train.
- Hassabis suggests DeepSeek’s cost figures might not include full development expenses and hints at possible reliance on Western AI models.
- Despite DeepSeek’s strong performance, Hassabis does not view it as a breakthrough, highlighting Google’s own Gemini models as more cost-efficient.