If vibe-coding is hopping into a self-driving car and telling it to take you anywhere you can get a coffee, then I use coding agents more like a bicycle - they let me get further faster than if I'd walked, but I still have to decide where to go and how to get there, and I still have to pedal.
I don't vibe-code, but I do decide what to implement and what patterns to use (perhaps asking the model to analyze and give advice on this first), then I have it handle the nitty-gritty of the implementation itself. For this usage style, the latest local models are as good as having Claude at home.
I won't say it's been _easy_ (I ended up implementing my own harness to accommodate the idiosyncrasies of local models), but I will say that for the effort, having a coding agent that's essentially free to query as much as I want has been life-changing as a dev, especially when it comes to working on side projects. Knowing that my agent will never get worse in quality, suddenly cost more than it does now, or be suddenly made unavailable by external factors, was absolutely worth the trouble. And on top of all that, I can't believe it's as good as it is.
It's strange to me that Apple would choose to disadvantage themselves by selecting Google as their provider as opposed to, say, Anthropic or even OpenAI. Doesn't this mean they'll struggle more to differentiate themselves from the assistant on Android phones? Thinking more cynically, couldn't Google, if they wanted, feed Apple an inferior version of Gemini, ensuring they stay ahead?
As the consumer, this just sucks because it means no matter which phone platform you choose, you're getting the same thing underneath, and there's no way to avoid it (besides not using an assistant entirely, which I recognize a lot of people do, myself included).
My Computer Engineering degree had an "ethics" course (really a course on "engineering communications", but it was considered to satisfy the ethics requirement for graduation). It was a semester on how to file memos, cargo-cult your resume, and tell recruiters what they wanted to hear. Not a word was said about considering the implications of the things you're hired to build. When defense contractors took over the entire ground floor of the engineering building to hold a recruiting fair, we were encouraged to go.
The only time ethics in engineering was ever mentioned to me was in a class on applied number theory (cryptography), taught by a professor who had previously worked for the EFF. He went off-topic to tell us that many problems, like how to hit a target with a missile, may fascinate and compel us as engineers, but we shouldn't let that distract us into building instruments of death.
That course was an elective, and it was entirely possible to complete my degree without hearing a single mention of ethics.
There are many reasons I look back on my academic experience with disdain, but this one stands out to me.
This is exactly why, as someone who thought I'd be an IC with my head buried in code my whole life, I accepted a role as a tech lead last year. Humans will always need other humans to be human for them. I love working with computers, but supporting, teaching, and mentoring junior engineers has been rewarding for me in ways that writing code never could be. There is no social substitute for concrete relationships with specific people that grow in visible ways. Maybe they can automate away the part of me that's good with logic and reason, but empathy can't be simulated.
Re: your edit - I've always thought it'd make for a fun pet project to take the hyper-engagement-driven approach that social media apps take but apply it to Wikipedia. Like a Tiktok-style vertical feed of articles that tries to find the ones you'll engage with most (WikiScroll [0] is already part of the way there). Or a chatroom corresponding to each article that people can banter in, kind of like you described. Or a DM feature but you can only send Wikipedia links. The idea of Wikipedia as a social catalyst is super interesting to me.
Contrary to the popular belief that Linux has no games, Arch includes many fun games such as "find out why the audio decided to stop working today for no reason"
I don't vibe-code, but I do decide what to implement and what patterns to use (perhaps asking the model to analyze and give advice on this first), then I have it handle the nitty-gritty of the implementation itself. For this usage style, the latest local models are as good as having Claude at home.
I won't say it's been _easy_ (I ended up implementing my own harness to accommodate the idiosyncrasies of local models), but I will say that for the effort, having a coding agent that's essentially free to query as much as I want has been life-changing as a dev, especially when it comes to working on side projects. Knowing that my agent will never get worse in quality, suddenly cost more than it does now, or be suddenly made unavailable by external factors, was absolutely worth the trouble. And on top of all that, I can't believe it's as good as it is.