“I want to add a command line option that auto generates helloworld.exe”

“That’ll be $21,000.”

If you were so smart you’d have wads of cash like them. They got where they are through sheer grit and bootstraps and a paltry $50 million from their family.

I agree with you on a technical level. I still think LLMs are transformative of the original text and if

when the number of sources that’s what ultimately created the volume of the N-dimensional probabilistic space they’re following is very low.

then the solution is to feed it even more relevant data. But I appreciate your perspective. I still disagree, but I respect your point of view.

I’ll give what you’ve written some more thought and maybe respond in greater depth later but I’m getting pulled away. Just wanted to say thanks for the detailed and thorough response.

This is interesting and the article makes this very clear up front but the title is a little clickbait-y, because this requires a fully compromised device. I think it should be fairly obvious that if your device is fully compromised that built in software safeguards are not reliable.

Thank you. Great addition. That was a very interesting read, though I need to be more awake for reading technical writing like that 🥱.

My point about spending $20k to produce garbage, then, was actually realized in this “perfect” use case.

Hey, so I started this comment to disagree with you and correct some common misunderstandings that I’ve been fighting against for years. Instead, as I was formulating my response, I realized you’re substantially right and I’ve been wrong — or at least my thinking was incomplete. I figured I’d mention because the common perception is arguing with strangers on the internet never accomplishes anything.

LLMs are not fundamentally the plagiarism machines everyone claims they are. If a model reproduces any substantial text verbatim, it’s because the LLM is overtrained on too small of a data set and the solution is, somewhat paradoxically, to feed it more relevant text. That has been the crux of my argument for years.

That being said, Anthropic and OpenAI aren’t just LLM models. They are backed by RAG pipelines which are verbatim text that gets inserted into the context when it is relevant to the task at hand. And that fact had been escaping my consideration until now. Thank you.

I just posted where I found the source in another comment. It would have probably the information you’re interested in.

Here is the original cite that my company pulled that from if you want more details.

I’ve never written a compiler, nor in Rust, so I have no idea the effort involved. I’m just boggling over the price tag. I’ll bet that’s the cost of an entire offshore team.

At work today we had a little presentation about Claude Cowork. And I learned someone used it to write a C (maybe C++?) compiler in Rust in two weeks at a cost of $20k and it passed 99% of whatever hell test suite they use for evaluating compilers. And I had a few thoughts.

• 99% pass rate? Maybe that’s super impressive because it’s a stress test, but if 1% of my code fails to compile I think I’d be in deep shit.
• 20k in two weeks is a heavy burn. Imagine if what it wrote was… garbage.
• “Write a compiler” is a complete project plan in three words. Find a business project that is that simple and I’ll show you software that is cheaper to buy than build. We are currently working on an authentication broker service at work and we’ve been doing architecture and trying to get everyone to agree on a design for 2 months. There are thousands of words devoted to just the high level stuff, plus complex flow diagrams.
• A compiler might be somewhat unique in the sense that there are literally thousands of test cases available - download a foss project and try to compile it. If it fails, figure out the bug and fix it. Repeat. The ERP that your boss wants you to stand up in a month has zero test coverage and is going to be chock full of bugs — if for no other reason than you haven’t thought through every single edge case and neither has the AI because lots of times those are business questions.
• There is not a single person who knows the code base well enough to troubleshoot any weird bugs and transient errors.

I think this is a cool thing in the abstract. But in reality, they cherry picked the best possible use case in the world and anyone expecting their custom project is going to go like this will be lighting huge piles of money on fire.

men & women
males & females
men & females

It does feel kinda weird, right?

I am antifa, and I’m completely disorganized.

Yeah. I’m pretty sure for profit social media isn’t good for anyone. Adults shouldn’t use it either. But we decided long ago that you can’t stop adults from drinking or smoking weed, so adults are just mature enough to handle it or approach lies and manipulation with more skepticism, but I look around and see it’s not true.

I feel like at least things like Lemmy and Mastodon are much easier to filter or walk away from when you aren’t in the right emotional space.

What we have now is “neat.” It’s freaking amazing it can do what it does. However it is not the AI from science fiction.

It’s close. Thinking something more specifically couples-based, though in general it could be more like that than match.com or whatever.

ChatGPT, write some AI prompts that only someone over 40 would ask.

• “Rewrite this email so it sounds firm, calm, and professional—but not like I’m mad, tired, or done with everyone’s nonsense (even though I am).”
• “Explain why my lower back hurts specifically today when I did nothing different yesterday.”
• “Summarize this meeting transcript and tell me which parts actually mattered and which parts were just people protecting territory.”
• “Give me a workout plan that assumes I used to be athletic, still think like I am, and will absolutely injure myself if I’m not careful.”
• “Create a dinner plan that’s healthy, not boring, doesn’t spike my cholesterol, and doesn’t require 14 ingredients or a personality shift.”
• “Help me decide if this friendship has run its course or if we’re just in different life phases.”
• “Rewrite this resume so it highlights experience without triggering age bias or sounding like I’ve been doing the same job since the Clinton administration.”
• “Explain this new trend to me, but also tell me whether I actually need to care.”
• “Draft a polite but unmistakable message that says: ‘This is not my problem, and it will not become my problem.’”
• “Compare these two life choices assuming I value stability, autonomy, sleep, and not starting over for the thrill of it.”
• “Help me articulate why I’m tired in a way that isn’t about sleep.”
• “Turn this rant into something I could safely say out loud without burning a bridge I might still need.”

There you go, kids. And also… yeah these feel pretty accurate.

AI is a much better tool to help experts than replace them.

Used how?

Like, Claude, write up an operational plan for capturing President Maduro.?

Or like, Claude, turn these crayon drawings into tactical plans.?

Or like, Claude, help me find Brazil on a map.?

You know what would be a valuable dating app to me? I have no idea how this would work in actuality, but an app that helps me date my wife in some way. Suggestions of what to do, packaged date nights, flowers, all that shit — except for all kinds of people not just “traditional romance.” River rafting, swinging, sports bars, I mean appeal to all kinds or even people who just want to try things they don’t even know is it’s for them or not. It would be great for singles, but would still be useful after a relationship is in full bloom. But I think that’s something completely different from a dating app.

“The highlight of my life is breaking 2k karma on a single 5 word shit comment.”

Inspired by a “true” story.

MD is a nearly ideal format. I keep my personal notes and time management stuff in Obsidian using markdown. Write my blog in Markdown. AsciiDoc is nice, too, for certain use cases.