Things read - Week 18

Using AI to understand something

This week started with thinking about AI and what kind of rewiring is needed to fully grasp AI in our professional and personal lives. I came across this post by Paras, where he outlines how he used Claude to understand GDP. He even shared the full transcript of his chat. I was hoping to find an extraordinary prompt or some other magic, but it was just pure, simple chat. My takeaway: more than a fancy prompt, it's the mindset that matters more. Challenge your assumptions and, as my friend says, assume that AI is going to do a better job, so start there and walk backwards from there. Link

High agency

Written by George Mack - basically needs to be revisited every six months or so. Such actionable advice that is also available for free is rare and precious. It won't be fair to summarise his writing in a paragraph, so I urge you to spend some time reading through it and marinating. I am sure something productive will come out of it. Having said that, I keep thinking about his suggested tool of Story Razor: when stuck between two potential choices, think about what would make the better story. We continually keep retelling our stories but rarely think about ourselves as the authors of the story. We consider ourselves as the characters of the stories. But having an interesting life story has a halo effect - it makes it easier to attract new friends as well as life and business partners. Link

AI connundrum

I was listening to this podcast recently, and found myself nodding in agreement that it somehow feels wrong to use these LLMs to build stuff. After all, writing code was the skill that we acquired after much tussle, and now we are handing off our skill to another program. If we don't exercise, won't we just lose our muscle and edge? But at the same time, using LLMs, we can build stuff at 100x speed, so should we really care about how that stuff is made? Shouldn't just the end goal matter? I would have never dared to make things that I have been able to make now. So I don't know. May we at least try to understand what exactly is going on under the hood and let LLMs build stuff at blazing fast speed while we think about WHAT to build. Link