Starting to wonder that we’re going to start being forced to execute on an AI output instead of sharing it with other people. If you can reason yourself into a working system, you know what you’re talking about. If not, then it’s not worth taking the time to figure it out.
The LMGTFY answer to questions we feel are simple is very strong. Sometimes I feel like we forget that we know valuable things which are valuable to others, and that’s still better than what AI will spit out.
It's not that there's a cost to generating them -- per say. I wouldn't want an LLM generating UUIDs anyways. I think it's the cost of consuming them in conversation context that is the issue.
This doesn’t seem serious. “Two years of access” when the models and harnesses just got good last November. It also requires people to want to spend the time and tokens to build something for which we already have free alternatives (which are good!) like the GIMP. People are building projects that are important and unique to them, not hypothetical complex software for the fun of it.
Maybe they themselves are fakers and liars / deeply insecure. I got bumped out of an interview rather rudely once because I blanked and couldn’t answer a trivia question about arrays.
It’s about deeply understanding what you’re doing. Like as a kid before you knew how to ride a bike, you could sit on a bike and peddling, but until it “clicked” you couldn’t balance and keep going forward stable. Fizzbuzz tests your ability to reason through a problem that seems simple on its face, but is easy to get wrong and/or overthink.
I would imagine that the strong type system would also help steer AIs to ensure that code is written correctly. The compiler supplies backpressure to the agent that steers it toward a more correct program.
This has not been my experience. Claude artifacts at first, then Claude Design after it was released, are excellent at design! The way I can steer the model updating the design with different ideas and visions, even adopting different design systems like Material 3 or Apple’s HIG it has been phenomenal.
Congrats to Chris Lattner and crew! Seems like a neat project. I listened to him talk about Mojo on some podcast a while back and was really impressed. Swift was a runaway success, and I hope he can pull it off again with Mojo! :)
This is great. We also need a tool to expose source jars to agents so they don’t need to compress. There’s a lot of Compose overloads that Claude just guesses at. I built something internally but it needs polish and Claude really struggled with the deep Gradle integration.
I had an idea like this a year ago. Super interested how it goes if this is real. The problem right now is that it’s hard to get your hands on hardware in the first place.
> When using a custom kernel collection with Apple Silicon, there are some unfortunate downsides. The biggest being that streamlined OS updates are no longer available.
Tart is an amazing tool, and I have been very grateful for it. There's almost no other way I've found to stand up ephemeral CI/CD macOS VMs for self-hosted Git forge solutions. I really hope this doesn't mean that Tart will eventually die the death of unmaintained projects (e.g. Realm post-Mongo-acquisition.)
Yes but most people are not dropping down to Metal support unless they're doing custom effects or developing a game engine. Most apps could be developed outside of Xcode just fine.
I suppose you would have to commit your node_modules, or otherwise cache your setup so that all prerequesite modules are built and ready to install without running post-install scripts?
It almost doesn't matter, because you can get pwned by a transitive dependency. If someone doesn't have the same scruples as you have, you're still at risk.
Part of the issue is that they don't actually advertise what the token limit is. Just some vague, "this is 5x more than free, and 5x more than pro". They seem to be free to change the basis however they please, because most of us are more than happy to use what they give us at the discounted subscription pricing.
How much do you want to bet me that the credential was stolen during the previous LiteLLM incident? At what point are we going to have to stop using these package managers because it's not secure? I've got to admit, it's got me nervous to use Python or Node.js these days, but it's really a universal problem.
Either god is me (secularism) or god is something outside me (Christianity). One is going to be better than the other. It matters which one. Everyone has an answer, and it affects your morals. Whether or not you are consistent brings you back to that same question: “who says?”