fwiw this problem already exists with my more junior co-workers. and also my own code that I write when exhausted!
if you have trusted processes for review and aren't always rushing out changes without triple checking your work (plus a review from another set of eyes), then I think you catch a lot of the subtler bugs that are emitted from an LLM.
there are a million million subcultures with pretty stark differences in taste/aesthetics that you can dig up on the internet. looking at what's grossing in mega-dense populations of millions of people then yeah, perhaps in aggregate at large N, things their individuality -- surprise?
there are parts of every major city that feel the same, but if you're willing to take a train out 45 minutes in any direction without google maps, i'm willing to bet you get into spaces that are incredibly local!
A new-ish field of "mechanistic interpretability" is trying to poke at weights and activations and find human-interpretable ideas w/in them. Making lots of progress lately, and there are some folks trying to apply ideas from the field to Alphafold 2. There are hopes of learning the ideas about biology/molecular interactions that the model has "discovered".
Perhaps we're in an early stage of Ted Chiang's story "The Evolution of Human Science", where AIs have largely taken over scientific research and a field of "meta-science" developed where humans translate AI research into more human-interpretable artifacts.
I'd upvote this twice if I could. Young male here and dating is rough; most of my friends are male and single...
But I find it a little ungenerous to interpret that young women are dating men who are cheating. I think it's more likely the case that single women who are in their 20s are willing to date men in their 30s, and so if you have to choose between a 23 year old who just graduated or a 33 year old well established in their career...
I like the sentiment of the title, even if it gets people riled up. I'm fondly reminded of the introduction from Bob Nystrom's _Crafting Interpreter_ [1]
> This last reason is hard for me to admit, because it’s so close to my heart. Ever since I learned to program as a kid, I felt there was something magical about languages. When I first tapped out BASIC programs one key at a time I couldn’t conceive how BASIC itself was made.
> Later, the mixture of awe and terror on my college friends’ faces when talking about their compilers class was enough to convince me language hackers were a different breed of human—some sort of wizards granted privileged access to arcane arts.
> It’s a charming image, but it has a darker side. I didn’t feel like a wizard, so I was left thinking I lacked some inborn quality necessary to join the cabal. Though I’ve been fascinated by languages ever since I doodled made-up keywords in my school notebook, it took me decades to muster the courage to try to really learn them. That “magical” quality, that sense of exclusivity, excluded me.
> And its practitioners don’t hesitate to play up this image. Two of the seminal texts on programming languages feature a dragon and a wizard on their covers.
> When I did finally start cobbling together my own little interpreters, I quickly learned that, of course, there is no magic at all. It’s just code, and the people who hack on languages are just people.
> There are a few techniques you don’t often encounter outside of languages, and some parts are a little difficult. But not more difficult than other obstacles you’ve overcome. My hope is that if you’ve felt intimidated by languages and this book helps you overcome that fear, maybe I’ll leave you just a tiny bit braver than you were before.
they have a blog hinting at some answers as to "why": https://turso.tech/blog/introducing-limbo-a-complete-rewrite...