the solution might be cancelling my AI subscription
the solution might be cancelling my AI subscription
I am trying to think of a list of all the wonderful things I've built with AI:
a speech recognition system in rust<br>an email archive rendering + quote collapsing tool<br>a jellyfin desktop clone with gstreamer and qt quick<br>an invidious clone in python + yt-dlp<br>a faithful Windows 95 notepad.exe clone in fltk ported from the Wine<br>sources<br>a machine vision thing to count traffic flows from public street cameras in<br>opencv<br>a claude ui clone in python or rust i think, i don't even remember<br>a regional news site i never meant to build that is actually getting traffic, python/flask<br>a 3d car game built on the protocol for an existing multiplayer game in<br>three.js<br>an investment backtester in python<br>a html clone of the lightroom ui, marvelled at the result then never made<br>the backend<br>a markdown viewer in qt or gtk or something else i can't even remember<br>a replacement world clock widget for my laptop desktop environment in gtk<br>and C<br>a javascript network synchronised audio playback thing<br>a rust client for a chinese IP camera reversed from its Android app<br>a sizeable SaaS in rust<br>maybe 50 other projects i've already deleted
Except for the SaaS, almost none of this is useful and I don't want to maintain<br>any of it. I accidentally run a news outlet which is surely a liability. Sure,<br>it has helped me "learn AI tooling" and I use many of these tools, but I didn't<br>need them. I can't afford to maintain any of them, not in terms of time,<br>commitment, belief, attention or willingness to spend on tokens.
I didn't mean to build most of these things. Usually the Claude session started<br>with something like write a quick script for X, and one hour later the<br>result is not a quick script for X, nor in the usual case is my<br>problem solved, whatever the original itch happened to be.
attention is all you need
On that last point, this technology is horrific for attention.<br>It's a thermonuclear ADHD amplifier and I have seen the same effect in every<br>single one of my adult friends. Folk running 3 screens simultaneously working<br>on totally unrelated "projects" they have little hope of maintaining, and such<br>little commitment to the outcome that the time is obviously wasted.
In recent times, at least once per month someone sends a screenshot for an<br>awesome tool they are working on. I'm like whoa, that's really something<br>and the sender is obviously proud and enthusiastic. I try not to ask, but am<br>always thinking and where will you market it?, because when the question<br>is asked of an engineer, the answer is unchanged since before LLMs existed.
I recently interviewed and when the topic of AI usage came up, the host<br>answered something like oh we're quite light on it, everyone has up to 5<br>rooms where they manage their agents and I immediately felt a tightness in<br>my stomach.
I had a vague sense of the effect a few months into using Claude. Later I<br>reduced my subscription to Pro in the belief a quota restriction would mitigate<br>excessive use. Then Claude went through a bad service period and I moved to<br>Codex. Codex's CLI is much nicer than Claude's and noticeably faster. And usage<br>started creeping back up.
The technology, when honed, is genuinely amazing. Ask it to zero shot a parser<br>for an esoteric grammar implemented in an esoteric language with full tests and<br>it's done. The tooling as it exists today promotes absolutely nothing like the<br>focus required to apply it judiciously.
Almost every vendor and every tool intends to do exactly the opposite: more<br>usage, more tokens, more output. Ask a simple yes/no question of ChatGPT and<br>you can clearly see that it is hard-wired to include a relevant follow-up<br>question to promote excessive interaction.
Slopping out a 10,000 LOC untested Python/JS mess in 5 minutes helps nobody.<br>The thought of this happening in every commercial environment simultaneously is<br>horrifying.
friction = focus, focus = product
One of my early AI experiments, exploring AI as a lens in Marshall<br>McLuhan-like thinking, was to connect speech recognition to a pipeline that<br>generated blog posts on the other side, in the belief it would encourage me to<br>capture my thoughts. All I needed was to press the voice note button in a<br>Telegram channel, and out pops an Opus-formatted post.
The output was unbridled garbage. Because the effort was removed, so was the<br>commitment, and with the commitment the focus, and with the focus any<br>meaningful product at all. Quality writing is not conversational English simply<br>cast through a lens: conversational English is low-bit rate noise, quality<br>writing attempts to capture high bit rate information with better formed<br>concepts, and this should have been obvious before I began.
I looked at repurposing the pipeline to capture private notes, but I have no<br>need for private notes. It subverts the natural process of noise being<br>forgotten. It is just more excess tool use.
Following from this, for as long as quality...