Instead of direct trivia from the content, it would be helpful to have an exercise (with evaluation) that applies the content learnt - a small artifact production with real-world practical use.
Imagine you would need, another ai pipeline that poses as the consumer and applier of the knowledge, instead of a direct processor ai of content information as it currently seems.
> Scale isn’t bad, at least not necessarily. Industrial is perfectly capable of being better than custom. Sometimes the YouTube video is more helpful than the private tutor.
This is not true; consider the argument that there is always a loss of quality in scaling. Industrial maybe better than average custom, but is always worse than best customs. Broadcast lecture is almost always worse than a tutor whose discourse is customised to student's current knowledge. (The word guru (literally, one who leads towards light) is wrongly translated to teacher, for which the word in sanskrit is shikshak.)
I agree the average tutor is better than the average lecturer. But if I watch a YouTube lecture, then I might have access to the best lecturer in the world, or at least a 99%ile lecturer, e.g 3blue1brown. This only works because of the scale -- it wouldn't be possible for millions to get access to tutors this good 1 on 1.
but can you argue with them for an indeterminate period of time in earnest dialogue until the idea goes CLICK inside your head? You cannot, and so the value is much lessened because that critical access isn’t present.
Do you always need that kind of prolonged argument whenever you learn a new concept?
> Sometimes the YouTube video is more helpful than the private tutor.
University lectures work like that. First the general lecture, which has 300 students and is not interactive. Then the tutorial, where students who want extra help can consult individually with a TA. Attendance for tutorials was typically much lower in my experience, because a lot of students thought the lecture was plenty and wanted to spend the time on something else.
Except the argument is university lecture (classroom structure) doesn't work. A personal tutor might be able to teach a concept to all their students, whereas not everyone in a classroom passes. Your argument about sufficiency can be made about books and is not relevant here.
> A personal tutor might be able to teach a concept to all their students, whereas not everyone in a classroom passes.
Why are you convinced that personal tutors are so much more effective? I'd expect the average student to be slightly better-served by having a personal tutor, but not by much. After all, personal guidance is available from TAs when necessary—my point is that it typically isn't necessary.
I don’t think that holds for everything. Industrial 3nm chips are probably better than whatever you would get if people tried to DIY this. Lots of things get better with scale: materials, precision machinery, process efficiency, power generation.
But there’s also plenty of things that don’t work this way. Care is one of the most extreme things that does not.
I think the argument is that someone with the knowledge to make 3nm chips could probably know how to make some very cool yet very expensive chips, but since they target a lot of people they probably make choices that would satisfy more people but maybe not as well
Something one might like as a continuation of the article is digital sundials. Apart from types listed on wikipedia, there are 3D-printed versions etc.
Its not that simple. Time it seems IS the first (0th?) dimension.
A point is space denotes existence in time of the object & observation by the subject. In other words, rate of change of existence is observed as time. Rate of change of a point is observed as a line. A moving point accepts line as its track, moving interval accepts square as its track and moving square accepts cube as its track. 2-eyed observer has 3D vision & 1-eyed observer as 2D vision (try touching your fingers with one-eyes closed exercise) has some role about role of observation as well.
Further, dimensions being relative vs absolute makes more sense. In absolute sense, time is its own dimension T & point line cube are L, L^2 & L^3. A 3D object, a cube, has 2D object, plane, as its boundary & 1D object, lines, as its dimensional denotion. A square has 1D object as its boundary & n-2=0D objects, points, as its dimensions, relatively speaking. This is important because of the number of eyes? So basically, those 2D hypothetical characters in your physics are 1-eyed creatures, lol.
> Prompt engineering doesn't feel like an activity that creates sustainable AI advancement.
Chatgpt was created from gpt via prompt engineering? An inverse chatgpt where user answers questions instead of the other way around also has applications.
Only in that the very initial version was GPT + prompt. ChatGPT is heavily fine-tuned using both handwritten examples and later on rated examples from its own output. The prompt is still relevant but the fine-tuning is the biggest thing that makes it work like a chatbot.
Binance or exchanges are 'anti' crypto. Buying/Selling crypto defeats the purpose of crypto entirely, rather supports fiat(do the math please). Exchanges must die before crypto even starts working.
We are still going to need on-ramps/off-ramps and it would be better if they were legal and regulated to be small. Exchanges are still going to be needed. "Retail banking" with managed wallets are still going to be needed. Only allow them to operate in one country, put a cap on the the maximum amount of funds/accounts they can have and get rid of any "investment" scheme.
Essentially, something like the crypto version of credit co-ops.
Honestly all I hope for is for people to stop thinking of crypto as stocks/investments but treat them as currency. Majority of the embarrassing press and downsides of crypto in the past decades have to do with the former.
Mining, swaping, burning, spending is enough. And where it's not, the use-case is not decentralized (has centralized governance or has known membership) & can be achieved more efficiently via fiat.
Nice tongue-in-cheek joke. Don't. At a given time, you are to be crypto savvy & be protected in the first place or don't be crypto savvy & use the legal system with fiat for better efficiency & solution
Imagine you would need, another ai pipeline that poses as the consumer and applier of the knowledge, instead of a direct processor ai of content information as it currently seems.