I check my credit card statements every once a while just to see if there are unrecognized costs. This month, I saw a whopping $600 bill from Google AI Studio. This was not cool. I knew I vibe coded things, but no way it could use this much tokens. Maybe something I built went viral? Maybe someone stole my API tokens for their own use?
Turns out, I was getting trolled by Google all the time.
A lawyer files a brief full of citations that don’t exist. A government relabels “patriotic AI” and points it at immigrants. A model behaves perfectly while it’s being watched - and only while it’s being watched. These aren’t the same problem, but they showcase various sides of AI safety risks. The thread connecting them is intention.
People like to talk about how simple things are.
In engineering, we use “simple” as the ultimate compliment - we want simple systems, simple code, simple deployments, simple everything. But lately, I become to realize that we could be talking about two completely different things when we say that word. And as we lean harder on AI to build things, we are accidentally optimizing for the kind of simplicity that might eventually make our systems, and even our brains, obsolete.
Vibe-code Golang one interface at a time.
During the holidays, I was working on a personal project in Go, and I wanted to use AI to help me do a few things (e.g. implement a downloader that downloads a file from Google Drive). However, I’m not a huge fan of having AI IDEs creating new directory structures or introducing abstractions that I need to read through and understand.
A lot of times I use GenAI to quickly prototype something like an app idea or a UI/UX mock for a site. I’d like this text-to-UI experience to be as fast as possible to quickly iterate.