@yongfook n a world where AI is shifting scarcity every day, nothing is the definite real moat.
The real moat is having a happy family with a healthy body.
@Huxpro Bros in frontier labs are saying even model inference acceleration is solved. I consider that to be the most backend-backend job.
So no, there is no real moat. Earn your last penny while you still can.
Agreed.
If you're worried about slops, then your loop isn't built right. That's the equivalent of asking a blind-folded engineer to build something and wishing it magically work.
I used to think that AI eval is something human need to curate, but now it seems that AI writes it better and faster than me. In which case, what is it I can still do besides just saying "Hey Claude/Codex please build an eval for this AI agent?"
1. replying.
2. replying to every point will make it a long reply.
3. this reply is taking $5 of my brain power as meat token cost.
4. this is not a lottery purchase.
5. most people will find this reply a fun read.
6. I will turn this into a post if it reaches 380 likes.
7. bookmarked before started reading this post.
8. nothing beats a casual X scroll and reply on a warm Sunday afternoon.
A technology whose externality reaches the whole world yet the governance can never be global - pacing AI has no practical chance.
Think about the Cold War.
At the core of our mission is working through how to ensure increasingly powerful AI benefits everyone.
We believe that, at some point in the future, AI acceleration for frontier model development may be so high that the world will need to pace the rate of AI advancement.
We hope to contribute to work led by the U.S. government, alongside other labs and the open-source community, to develop the tools and mechanisms that could make that possible.
https://t.co/pMCtiQjMoo