Remix.run Logo
▲ augment_me 3 hours ago

This is a take that is outdated if not now, then in 3 months.

These has been this ladder of copium that "oh human still has input here for X, oh shoot it doesn't with Claude 4.5, well human is still needed for Y, oh shoot...".

You are suggesting that you can provide context that the model can't on its own. This is a search problem, every iteration it gets better.

We are at the stage in our company where "/goal optimize this" gets better result than any biased human context engineering can provide.

▲zahlman 2 hours ago | parent | next [-]

> You are suggesting that you can provide context that the model can't on its own. This is a search problem, every iteration it gets better.

How exactly do you propose that the model should search my thoughts? Of course I can provide context that the model doesn't have. I can provide context that does not exist anywhere on the Internet (unless I got hacked).

> We are at the stage in our company where "/goal optimize this" gets better result than any biased human context engineering can provide.

First, a "this" must exist to optimize. Second, you have to be happy enough with it that optimization is the only problem.

▲shmolyneaux 3 hours ago | parent | prev [-]

I think it's reasonable that humans need to provide context for this, and situations like this, forever. The GP has a specific problem. The context of the problem exists in the real world, not something that the agent can search for. This isn't a "humans are still important" take. This is custom software for meeting the specific needs and preferences of an individual.

I fully agree with them that the easiest solution is to ask an agent for exactly what you want, and iterate from there. Evaluating and installing and learning off-the-shelf software is often more of a pain than doing it from scratch.

▲augment_me 2 hours ago | parent [-]

What prevents a solution where you physically hook up the needed components and then type "/goal make this keyboard extension work"

And the same library is built without the "seed" or baseline?

▲zahlman 2 hours ago | parent [-]

What does "work" mean? How should the agent know what UI you want? How should you know, until you try something out?

▲augment_me an hour ago | parent [-]

Answering here for both of your comments:

1) If the UX/UI is important for you, the agent can have access to your other work, or to your other conversation and build your tool with that in mind. This is just another search space question. If you have a documented, predictable taste, the models can efficiently produce what you need.

My uni supervisor/professor used to complain about my slop papers, then I started generating them in his writing style by just pulling down all of his papers and blogs, and he doesn't mind it now.

2) In general, the LLMs are great for steering opinion. There is good research on steering political opinion with LLMs. Your taste and opinions can be formed by the LLMs to like the solution that it provides you.