Even if LLMs were trained on the answer, that doesn't mean they'll ever recommend it. Regardless of how accurate it may be. LLMs are black box next token predictors and that's part of the issue.
Complications and Being Mortal by Atul Gawande. Made me appreciate modern medicine more. One other l book that shifted my thinking completely about AI and how far we are from AGI: A Brief History Of Intelligence by Max Bennett. Evolution is a heck of an algorithm.
Also read Apple In China. Was pretty interesting to realize how much Apple (and China) are what they are because of how much they poured into each other
I'm building a utility to help DJs find "play-out" versions of tracks they already like[1]. You can play with it here[2]. Streaming services are optimized for Radio Edits. But to actually mix a track, I usually need the Extended Mix, Club Edit, or a specific Remix. Manually searching for the "DJ version" of every single track in a 50-song playlist is tedious administrative work that kills the joy of digging.
Remixify automates the search while leaving the selection to you. You paste a Spotify playlist URL, and it helps you or provides you a good starting point for digging. It groups the results by the original track so you can quickly preview and save the versions you want to a new playlist.
We don't try to recommend new music or use AI to guess your taste. It just finds the usable versions of the music you already selected.
I resonate highly with this. Especially when brainstorming ideas with my manager. He's very quick with suggestions, and I am always saying ehhh I don't know let me think about it. I have realized that him giving me ideas quickly to iterate on is beneficial because I am always able to refine it.
I still do think it is a deficiency in some sense as I would have loved to be one of those guys who could just grok stuff instantly and contribute quicker
Let me give you guys a perspective from someone who did Electrical Engineering in Africa. There are hardly any job openings centered around electronics so the programme/curriculum is Power Engineering heavy. Most of the professors did research around this area, and there were very few who did anything in the sciences. I had the chance to pick some electives from the Computer Engineering department, but this was just during my third and final year of university. Unlike OP, I did not have an issue breezing through circuit problem sets, though they felt very repetitive, I was not bad at it, nor did I fail any classes.
I had an eye opening experience when I had my first taste of programming when I took C programming in my second year of university. What do you mean I can run a command and see instant output? Amazing! This was not the case for my electronics and power engineering lab sessions. We were using equipment that had been around since the 80s with little to no supervision. Just a bunch of routine "experiments" which I can barely remember any of. In my third year, I took Digital Computer Design (a C.E elective) and I realized I had been wasting my time learning about how the power grid in my country works. I tried my best to salvage as much as I could by picking more C.E electives, albeit not many available, did as best I could.
Everyday I wonder, how different would my life have been if I studied CS or even CE, I do not know. But, I appreciate the little this journey taught me, that you can always squeeze lemonade out of whatever lemons life gives you. I see my old EE notes now and they don't make sense to me, but I appreciate the happy chills solving circuit problem sets gave me. I work in software now, and I get that 1000x more, and that is how I know I made the right choice.
I don't think we will get to a point where we can safely mitigate the risks associated with this. It is almost futile to pull this off at scale, and the so called "benefits" are not worth the tradeoff.
reply