Is this like learning calligraphy in the typesetting era?

Before the AI era, I didn’t know much bash, but I was a reasonably OK programmer besides that I think. I found by getting AI to write me a lot of bash scripts and following along and then making edits myself when I needed small things changed I ended up with the ability to write bash now, and actually kind of appreciated as a language where as before I thought it was confusing. YMMV

Like anything with enough dedication you can achieve what you want.

▲

Antibabelic 5 hours ago | parent [-]

This is a strange analogy, because learning calligraphy is essential for any type designer worth their salt. Read The Stroke by Gerrit Noordzij.

▲

keepamovin 4 hours ago | parent [-]

I don’t mean type designer I mean, the Gutenberg press. Before mechanical printing books were copied by monks using calligraphy weren’t they?

▲

jraph 3 hours ago | parent [-]

It's not exactly the same thing.

When the Gutenberg press exists, knowing how to copy whole books by hand is 0% useful anymore, including to run a book copy using a press. There's also virtually no advantage to hand copy a book when you have a press.

You still need to know how to program to build something and maintain it in the long run. You need to be able to understand the Gen AI's output or you are in for some trouble, and to deeply understand the Gen AI's output you need to have practiced programming. What's more, you need to have practiced not only (generic) programming, but the specific stuff you are working on (the domain, the specific technologies, the specific codebase).

▲

keepamovin 3 hours ago | parent [-]

It was a little bit of a humorous tease however, I think there’s a side to it you’re missing as valid is what you say right now is.

	▲	jraph 3 hours ago \| parent [-]
		> It was a little bit of a humorous tease Whoops, missed that, sorry for this! Not sure I understand the rest of your sentence, I understand that you are saying what I'm saying is only valid right now but could change as Gen AI keep improving. I personally think this stuff significantly improving will require a breakthrough / paradigm shift, and that the "sophisticated stochastic parrot" model, even with "reasoning" stuff "patched" on top might only go so far and might quickly plateau (this is not science, only mostly uninformed opinion though).