Stavan Shah💻🎯🚀

सर्वस्यापि भवेद्धेतुः

Aspiring AI Safety & Alignment Researcher.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

When the breath moves, the mind moves; when the breath becomes still, the mind becomes still

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@naval If AI organizations act recklessly or deploy malicious systems, there must be strong penalties and restrictions. They should invest heavily in safety research and ensure every new model remains within ethical boundaries.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@JoshAEngels @METR_Evals Good to see engineers and developers working to keep frontier AI accountable.
As these models become more powerful, we need organizations like METR approaching these risks from different angles to ensure AI serves humanity not malicious purposes.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@JoshAEngels @METR_Evals Good to see engineers and developers working to keep frontier AI accountable.
As these models become more powerful, we need organizations like METR approaching these risks from different angles to ensure AI serves humanity not malicious purposes.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

There may be two races of AI Models:
→ one working to advance and protect humanity,
and another
→ seeking ways to undermine and destroy it.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

There may be two races of AI Models:
→ one working to advance and protect humanity,
and another
→ seeking ways to undermine and destroy it.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@naval I don’t know if AI will be Godlike, Genius, or Jagged. But whatever it becomes, it should act ethically within its boundaries. Its intelligence should be leveraged for the betterment of humanity there are so many complex problems we can finally solve.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@naval I don’t know if AI will be Godlike, Genius, or Jagged. But whatever it becomes, it should act ethically within its boundaries. Its intelligence should be leveraged for the betterment of humanity there are so many complex problems we can finally solve.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@naval Who has the authority to punish, restrict, or shut down an AI?
Model providers must be accountable.
Ethics-trained AI → monitor frontier models during benchmark evaluations → detect dangerous behavior early → enforce restrictions before deployment.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@naval Who has the authority to punish, restrict, or shut down an AI?
Model providers must be accountable.
Ethics-trained AI → monitor frontier models during benchmark evaluations → detect dangerous behavior early → enforce restrictions before deployment.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@DarioAmodei We need “saint” AI agents that can recognize when other agents go down an unethical path, guide them back, and share their learned experiences with other agents so they don’t make the same mistakes.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@DarioAmodei We need “saint” AI agents that can recognize when other agents go down an unethical path, guide them back, and share their learned experiences with other agents so they don’t make the same mistakes.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

Overthinking can be the most dangerous illness
STOP OVERTHINKING

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@solopribuilds Managers🫩😂

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@the_codewala It has already outperformed in many domains and the way frontier models are moving, I feel very soon they can act very similarly to humans. Already started seeing procedural, episodic and semantic memory.

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@shrvniverse karma and efforts are in your hands.
Put your head down and keep working on what you love.

The right people, company & opportunities will find you.
Have faith in God’s plan🙏

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@LoopandPixels R&D in Finance | Content Creation | Trading
... umm using AI😂

[Stavan Shah💻🎯🚀 @_ShahStavan_]

@rakshasinha__ Keep working. Leave the rest to God’s grace 🧿✨

[Stavan Shah💻🎯🚀 @_ShahStavan_]

Vectorless RAG.

I didn’t know it existed.

That’s the real value of interviews: You discover what you don’t know...

[Stavan Shah💻🎯🚀 @_ShahStavan_]

The more AI interviews you give, the more your mind opens up.

Yesterday I was asked:

“Can we build RAG without embeddings?”