There is also the matter of how training data was licensed to create these models. Local or not, it’s still based on stolen content. And really what transformative use case is there to have AI in the browser - none of the ones currently available step outside gimmicks that quickly get old and don’t really add value.
You watch too many movies, there are plenty of other things for the missiles to track. Transponder in civilian airspace is just how you keep planes from crashing into each other.
I don't watch those kinds of movies. I have, on the other hand, worked for a large aerospace firm supplying these weapon systems.
A transponder is how civilian planes tell exactly where they are relative to each other. Missiles use IR, radar-bounce, and other methods for the last-mile delivery of explode-y bits, but when launched from afar (e.g. a surface-to-air missile launched from land towards something over the horizon) they need to be pre-loaded with the exact position of the target, as they need to get close to it before switching to a homing mechanism. If the target has a radio transponder, that makes this step trivially easy.
If Venezuela wanted to shoot one of these planes down, with the transponder off the missile is as likely to lock on to a commercial airliner. They're not going to take that risk.
You’re looking at this from a point where the only piece of information about you out there is the data collected by the roomba. In reality, every sensible data broker would just add that signal to your already verbose profile and feed it to a model to determine the stuff you’re likely to buy… or would trigger you to generate engagement or whatever is needed.
The privacy danger here is not the one data point, it’s the unknown amount of other parties who will mix and match it with more data.
With GDPR, I’ve been requesting copies of my telemetry from various corps and it’s amazing the kind of stuff they collect. Did you know kindle records very time you tap on the screen (even outside buttons), in addition to what you read and highlight and pages you spend time on? Now add to that your smart tv’s insights about you and your robot vacuum cleaner … you see now this all grows out of control.
Claude reads and writes Gleam just fine. I think as long as the language syntax is well documented (with examples) and has meaningful diagnostics, LLMs can be useful. Gleam has both brilliant docs and diagnostics rivalling Rust. Gleam is also very well designed language wise, not many reserved words, very explicit APIs… also something that helps LLMs.
Contrast with the likes of Swift - been around for years but it’s so bloated and obscure that coding agents (not just humans) have problems using it fully.
You mean the government that’s already the tax authority for which you create and report invoices? I can promise you it’s a good thing. Electronic invoices are a lot easier for us (obviously since it’s just a click instead of a whole PDF operation). It also removes a whole bunch of possibilities for mistakes.
> you have to pay to see the entire standard
The article is wrong about that though, all technical docs are available via the europa portal and reference implementation examples.
If there are so many React developers out there using server side components while not familiar with the concept of CVEs, we’re in very serious trouble.
So people have different definitions of the word, but originally Vibe Coding meant "don't even look at the code".
If you're actually making sure it's legit, it's not vibe coding anymore. It's just... Backseat Coding? ;)
There's a level below that I call Power Coding (like power armor) where you're using a very fast model interactively to make many very small edits. So you're still doing the conceptual work of programming, but outsourcing the plumbing (LLM handles details of syntax and stdlib).
I know tech bros like to come up with fancy words to make trivial things sounds fancy but as long as it’s a slop out process, it’s vibe coding. If you’re fixing what a bot spits out, should be a different word … something painful that could’ve been avoided?
Also, we’re both “people in tech”, we know LLMs can’t conceptualise beyond finding the closest collection of tokens rhyming with your prompt/code. Doesn’t mean it’s good or even correct. So that’s why it’s vibe coding.
My meaning was that if we actually decide on definitions that make sense, for the specific things that different people are already doing, then there will be a lot less confusion on the matter!
The original definition was very different. The main thing with vibe coding is that you don't care about the code. You don't even look at the code. You prompt, test that you got what you wanted, and move on. You can absolutely use cc to vibe code. But you can also use it to ... code based on prompts. Or specs. Or docs. Or whatever else. The difference is if you want / care to look at the code or not.
It sure doesn't feel like it given how closely I have to babysit Claude Code lest I don't recognize the code after Claude Code is done with it when left to its own devices for a minute.
No, that's not the definition of "vibe coding". Vibe coding is letting the model do whatever without reviewing it and not understanding the architecture. This was the original definition and still is.
reply