Before LLMs, learning on the job looked like reading documentation. Now it’s a guided tour with verification. When I produce things in this way, I’m not just blindly accepting it. The goal is that by the end of it I have learned more about the codebase and architecture, not less. I feel that’s important.

Many people don't understand this, even big tech engineers. They see LLMs as a bottleneck. It's more that they don't understand how to use it to multiply their skills, just basics and code gen.

I use multiple Claudes at a time, daily. It's precisely because of that experience that I wrote:

  LLM + framework you don't understand goes in ... unmaintainable garbage comes out.

Claude follows code patterns and structure. If you setup that structure and those patterns properly, it will produce great code. If not, it will follow ... whatever it feels like, with each commit.

If you just have it built something with a framework you don't understand, it will do so just fine! But over time every "vibe coded" change you make will drift it further and further, until you are left with a mess of vibe-coded spaghetti.

Oh joy. A model whose safeguards make it prone towards code that make your systems less safe. How brilliant!

Yes, and "in collaboration with the U.S. Government" feels like a very gross ploy at appeal to authority. You don't need Mythos or really any SotA frontier model to make malware or do extensive penetration testing/reconnaissance already. Sure, Mythos might be faster/more efficient, but the cat has been out of the bag for awhile. Even the terminology "infrastructure providers" practically screams "Enterprise leads".

I don’t think you’re going to find a consensus on this because it really just comes down to the quality of the employees in each discipline. Actions speak louder than words. I’ve seen the IT people GP is describing. I’ve also seen yours. In GP’s scenario, they often even mean well but are very overwhelmed because they’ve come to exist in a space where _everything is IT_ because no one else is remotely qualified to fill the specialty gap. When I found myself in your scenario, the opposite was true and it completely matched what you described.

Anthropic (re-?)confirmed this.

It works until they get to the sentience part. Neat idea!

Even there it works a bit.

> These models are the only other things we've ever met that can hold a conversation, and they're made out of weights

Is a fair point.

Not especially. Depending on where you set your standards for "holding a conversation" you can satisfy the requirement with a classical markov chatterbot, a well-trained parrot, a copy of Eliza, or a telemarketer flowchart drawn on a sheet of paper. Only the markov bot is made out of "weights" in the sense of a statistical model.

Parrots are intelligent animals, albeit with a limited capacity for vocabulary and syntax compared to a human, and Eliza and the flowchart are made out of explicitly encoded rules and conversational tactics.

The quality of "conversation" you can have with everything on your list is highly limited, and is categorically different than the sort of conversation you are able to have with any modern AI.

"I am more comfortable speaking to an LLM than a person" is something that should make you reassess yourself, not dismiss the rest of humanity.

Those numbers are bullshit. The bottom of a grown woman’s cycle is around 100 pg/ml. I have no idea where they’re getting those numbers but I assure you they’re wrong, and even if they were remotely correct about E levels being close to one another at some point in the cycle the majority of the time they’re much farther apart.

Gold is pretty soft. You would have to cut it to 10 carat, so there’s be even more to go around!

Nah literally crushing plates would feel so good. Worth the effort to melt it again every few sessions

You could encase them in plastic to prevent damage and mask them for some run off the mill equipment. Nobody would suspect anything without prior knowledge.

> Nobody would suspect anything without prior knowledge.

The weights being nearly twice as heavy for their expected size would be a bit of a giveaway to anyone who has ever been in a gym.

That's easily sorted by higher plastic to gold ratio.

But they feel nice to touch.

Having to handle the plates with care and the damage they’d take regardless would add to the charm.

This whole thread renews my faith in humanity

I’ll spot you!

It's practically karmic how rich this is.

Right. It's deterministic, and determinism should be the goal. It's not metaphysical. Some users know what they want while others do not. The software we create (by any means) should give users who know what they want the tools to find it, and guide those who don't until they do. Software exists to help us create our fate. It surprises me how many people are willing to relinquish that control or never wanted it, even within our ranks, by using AI to simplify experiences. IMHO, the optimization for most, but not perhaps not all, tools is to introduce AI internally to refine, create and expose more parameters, not less. Search is a perfect example of this.