This year I’d like to write more. Exactly what I’ll be writing up about is a work in progress (ideas and feedback are very welcome!), but I endeavour to make it interesting and regular.
So without further ado, here’s a few cool things I’ve learnt this past week:
Love languages
I’ve always thought that love languages were limited. That they are reductive and a poor helpful tool to understand people.
My opposition was because any natural language has two skills: a speaking skill and a listening skill. And so why shouldn’t the language of love? In particular, the way you like to give love might be different to the way you like to receive love.
So I was delighted when the founder of dimensional.me, a personality fingerprinting app, investigated exactly this. The whole post is fascinating (it’s short, so if you’re interested I recommend you read the whole thing), but one of my favourite takeaways is that only 14% of people give love the way they want to receive it. Something we should note, when trying to better understand someone.
Ancient Mathematics
The Babylonians,an ancient civilisation located in modern day Iraq, were able to approximate √2 to 99.999% accuracy 🤯 And the mechanism they used is ingenious:
Without using search and a calculator or a computer, I struggle to figure out a way I could do so myself.
This makes me wonder - to what extent do constraints engender innovation? What constraints could I apply to my own life to drive deep thinking and focus?
Williamsburg Curiosities
Whilst cycling around Brooklyn, I realised there’s a gap in Citibike stations around South Williamsburg:
How odd!
I looked into this. Apparently it’s because of the impact of scantily clad women cycling past the Hasidic Jewish children - go figure!
What comes after ChatGPT?
Finally, a few comments on ChatGPT. ChatGPT has been out for just over a month now. We’ve seen people use it to simulate a virtual machine, fix bad scanning of text, and act as a programming assistant.
But one of the most interesting parts for me is how ChatGPT and other simple models could impact the workplace of the future. I enjoyed Benedict Evans’ angle — to consider where ChatGPT could be useful, given that it can provide directionally accurate but specifically wrong answers. It wouldn’t be helpful to write a new encyclopedia, for example, but could be powerful programming assistant to a human, who can verify and correct it’s output.
Noah Smith and roon added a great perspective to the conversation. The takeaway of their piece is that AI won’t replace jobs, but it’ll replace tasks. Humans will oversee the output of AI and will focus on what they’re uniquely better at — that they have comparative advantage on.
That’s it for now - catch you next week!