Alexandr Wang Says the Rule for AI Is Simple: "Scale Everything"

Alexandr Wang says the rule for AI is simple: "scale everything."
He thinks the biggest gains come from scaling multiple dimensions together rather than pushing one in isolation:
- Scaling a single part of an AI system can work, but it is not the optimal approach
- Performance improves much more when several dimensions scale at the same time
- His broad takeaway for AI development is simple: scale everything together