Everyone I have seen in a corporate environment is using it to do their existing work faster. The rest of the organization has not yet evolved to take advantage of these new tools.
Not saying I enjoy it. Not saying that I’d drive a giant banana either. But if I saw a police car, pull over a giant banana, I would think it was hilarious. That’s the definition of whimsy.
I have two sessions going. One is fine, one keeps timing out. Both Opus 4.8 in Claude code in terminal. Must have them routed to different to different infra that isn’t equally impacted.
He's calling the adventure "The World Needs More Whimsy Grand Tour." Sometimes it's ok to have fun. Nobody drives a big banana thinking they're not going to attract attention. It's part of the fun and whimsy.
You can’t read the headlines, they’re marketing like anything else. Companies are free to spin them however they want and journalists are happy to write them however they will scare you the most. Those are the only truths.
The recent AI craze is really just LLMs. I feel like finance was likely already the most AI-adopted industry without LLMs and their impact the last few years may have taken them from “80 to 100” where most industries are going “10 to 50.” Go back to 2022 for a moment and I think this article is identical.
A very intentional line in my CLAUDE/AGENTS is to act as the Principle Engineer and to explain architectural decisions along the way. I'm not about blindly generating the outcome. I'm reviewing decisions, asking deep questions, performing research and basically taking a college-level course in a specific topic with every big project I deliver.
For every skill that may atrophy, I feel like I am developing experience in 10 new ones. I am not focused on using AI to perform my existing job function though, I am developing new capabilities.
Know how they grew Xai and built the data centers so fast? They rerouted Tesla chip orders. That business was supposed to be built inside of Tesla, but it was extracted. Tesla still needs it.
Even better, thanks for clarifying. It’s that kind of omission from the article that makes the rest of it hard to swallow. Even if it is technically correct. Which is sadly the case for most “journalism” these days.
GPS was always a dual use system. This is very detailed and specific, but not interesting or surprising. Research has been study GPS signal data, found parts that are encrypted and he doesn’t understand. The end. Article seems only intended to generate an emotional response of “how dare they use GPS for war, man!”
They’re amazing too. That’s my point. The legacy launch providers were doing zero innovation, limping along shuttle era designs and literally buying the most critical parts from our biggest competitor. The people who hate Elon have no concept how revolutionary the Merlin engine was given this context. It doesn’t matter if SpaceX is successful or not, they revived the entire US space industry. That’s what matters.
And to think, it wasn’t that long ago competitors we still using old Russian engines for their domestic rockets. Brilliant work to get back to leadership in this domain.
Wow, similar sentiments about this being a throw back. I’d rather roll my own almost everything these days, may not be as good, but certainly won’t be targeted exploited broadly.
There are so many individual features in this program that have never been done or even attempted before. That's "Advanced" in my book. Yes, they attached it to an overly ambitious program that is rife with delays (and hubris) but the program started on its own, is the best path to making the 2028 landing happen (it won't), and on its own is incredible.
It's so much further advanced than anything anyone else is working on, does it really need to be on schedule? I feel like "on schedule" only pertains to non-research-intensive projects.
I like this abstraction. If the baker says “I could sell 10x more if only I had shoes that allowed me to bake faster” then the cobbler says, “split the growth with me and I’ll craft you all the shoes you want.”
I know insurance for a launch is typical, but seems really tough to do that for this still “rather experimental” launch. I got to imagine it has costs something like 50% on a project like this.
You have a great process. I'm doing something similar, fine with Tailwind as a base and just build my own components from there. I think this will help entry-level people develop a similar process so perhaps it's great for that.
Slop code is like an early neural net. Path of least resistance. Except we cna see it and compare it to how it’s replacing something we’ve traditionally done instead of a neural net being opaque. As they write more code the path and concern for how it gets there will be approach zero.