(if you don't care about my yap and just want to see my progress learning go, skip all this and just go to the first heading)
recently, anthropic published new research examining the impact of AI on how people form new coding skills: https://www.anthropic.com/research/AI-assistance-coding-skills
a close friend and i have already had our suspicions about this for a few months now, and we've been building some stuff that closely ties into this based on our beliefs (i'll likely be sharing that at some later point in the future).
separately, reading this article got me really curious, since it's frankly been a really long time since i lasted coded without any AI assistance whatsoever. how much AI i'm using is highly dependent on the task i'm working on (and how much it matters LOL), ranging from ideating and deciding on all the logic and using AI as the "hands" to write the syntax more quickly (for highly important tasks / anything work related), to sometimes vibe coding if i'm just doing something for fun and won't be publishing it. but in general, my coding workflow these days has become to spin up a claude code instance whenever i sit down to begin working on something.
i was really curious to see if the coding progress i've made in the last few years would actually hold up if we strip away the AI. as a reminder, i'm pretty uniquely situated (along with most other people my age), in that AI was not at all prevalent early on when i learned foundational cs concepts, and started rising in prevalence as i got a bit deeper into my cs education. so, my personal growth as a developer has largely coincided with AI's growing usefulness in writing code.
to put this to the test, i decided that i'm going to speedrun teaching myself any entirely new language that i've never written a single line of code in (go), entirely without the use of AI. i'll document my progress for around a week or two, and then see if i can: 1. achieve a similar level of output as my usual tech stack (typescript, usually nextjs) with go when i switch back to my normal AI-assisted workflow, and 2. try to then pick up another entirely new language (i'm thinking either rust or ruby) without doing the due diligence of learning it first, and see how my competency with go compares to my competency with that language, if i use AI to work with both.
my hunch is that honestly, this is going to go really well, and the results will be significant. i'm fairly confident that i firmly dodged the wave of over-reliance on AI when first learning cs first principles by ~ a year, and the experience i've gained with other languages is deep enough at this point that it will transfer over relatively quickly. the results i'm expecting to see are: 1. i will probably be able to reach a relatively productive level of familiarity with go quite quickly, and 2. i'm going to completely shit the bed when i try to code with rust using only AI, thus reaffirming my friend and i's hypothesis that understanding of core cs principles and true developer experience are still necessary to build anything technically meaningful, AI or not.
(hopefully this doesn't go terribly and i have to eat my words later LOL)
day 1 (feb 4, 2026)
day 1 started off pretty standard. i went through the go documentation and followed their hands-on introduction with the language, which walked me through running my first program (what was essentially hello world with some modifications), creating a second go module and using it in my hello module, and writing some tests. it was pretty standard stuff
after that, i decided to run through some easy leetcode problems to try to lock in the syntax. i did two problems (two sum and merge strings alternately) and then called it a day.
on the language front, i'm already liking go. the things that stand out to me are that it's pretty opinionated in a way that forces you to follow best practices (or at least what they would like best practices to be). for example, mandating that exported functions start with an uppercase, or that an else must be on the same line as the closing curly bracket of the preceding if.
this is in pretty stark contrast to my glorious king typescript, which is simultaneously my favorite and least favorite thing about it (the ability to write absolute slop / the most horrible code known to mankind that can still compile). i think it's a breath of fresh air because of how used to typescript i am and overall i like it.
i also love the built in support for testing. i think its genius how easy / intuitive it is to write unit tests out of the box.
the dependency management and module/package structure also felt relatively intuitive to me and i was able to wrap my head around it pretty quickly.
i think things are going really well so far. i'm pretty sure that i can hit the ground running and get started on actually building stuff out pretty quickly from here.