The idea hit me when I realized I parent AI agents the same way I parent my son. Not by doing the work for them. By watching, stepping in at ambiguity, and knowing when something feels off before it breaks. Would love to hear from other builders are you experiencing this too?
3
105
Sort replies: Relevant Recent Liked
Replying to @AvinashDalvi_
Very interesting take!
1
1
86
Thanks Rohit.
28
Replying to @AvinashDalvi_
This is why I feel like we are going down the wrong path with LLM's looking for AGI... AGI needs to be born and educated within a sandbox... just like a child.
1
2
30
The sandbox idea is interesting. But who defines the boundaries of that sandbox? That is the harder question.
1
1
19
Replying to @AvinashDalvi_
Love this analogy, @AvinashDalvi_! 🙌 The real skill isn't prompting AI better. It's knowing when its confidence is misplaced. That only comes from years of scar tissue, as you perfectly put it. The future is Human + AI, not Human vs. AI. 💯
1
1
19
Thanks 🙏🏻
16
Replying to @AvinashDalvi_
As a new parent I am so tuned in to this analogy! Thanks Avinash, that was a fun read! Funny thought, My son’s context window is sooo small, he’s still learning to predict next word/token 😄
1
1
22
haha I love that context window analogy. Small model and large model 😂
1
1
23
more replies
Replying to @AvinashDalvi_
The biggest danger may not be AI acting alone, but us becoming too comfortable to check.
1
1
16
exactly. the parent who stops paying attention is the real risk. not the child acting independently. comfort is where oversight quietly disappears.
4
Replying to @AvinashDalvi_
Very well put @AvinashDalvi_ Worth reading!
1
1
32
Thanks for reading.
1
28