dingaling911

For sure LLMs can be integrated into applications but at the level of MCP. A well-defined surface to expose functionality.

The frightening thing is how easy it is to make copies of things that can reason without the requisite investment, and how easy it could be to direct them to bad things.

> Maybe she was still in her party age/too young?

It was the other way around, she never partied. Nice gal.

Story of my life.

Hypergamy is real. "I need to have a baby, but eew this guy is a dork, I'm not giving him my womb, and look at all these cool guys on insta" is a real conversation I've had.

Had one drop me for a professional DJ.

A friend of mine was convinced she was going to marry a famous basketball player.

You can guess how these "relationships" went.

Super average women with one or two OUTSTANDING characteristics (get it.)

It's a real problem.

Fortunately, I already had kids when we were both younger and dumber which might ultimately be the solution: keep humans off social media until they're grandparents.

Very cool but I'm not sure I understand why a specific service is needed to allow agents to communicate. Why not just email or chat?

Structured, deep and wide

I might be dumb, but I don't quite understand why just running stuff in sub agents isn't sufficient for the context problem.

Eh that’s about medical diagnoses. Not directly transferable. For regular usage, multiple models just make you feel productive but I bet they aren’t any more productive than just one.

Looks very cool.

Now all you have to do is run NSL under WSL.

There is zero evidence for this "use multiple models" trope, IMO.

It's just feelings.

What do you think about this post from DHH where he has recently said writing code by hand is over?

What is Zawinski up to these days. Havent heard that name in a while

Have never heard of a modern professional army explicitly targeting children before.

Come on man. Give it up. These are cruel people.

So the conspiracy is that Israel killed its own people, but in a different way than the person said.

Brother, it's literally from their own news.

I can only imagine a GPDR-type regulation with ridiculous fines for rogue agents. Black or white access will restrain creativity and potential.

Just FYI, regulations don't prevent murder.

There needs to be a technological solution.

looks like this https://github.com/rfonseca/xkcd-gnuplot

I think what you're seeing is A LOT of moving fast that wasn't possible before. When you move fast, you break things.

Where it would take 6 hours to make a first cut of a static site, you can now do that in 15 minutes. Obviously that's not going to be perfect.

please upvote then, people are downvoting because they don't like that it works.

This problem is easily solved. "Check that the blog looks the same in XYZ browsers".

I have LLMs align stuff for me all the time because I'm too lazy. It takes the screenshot, changes the code, code auto reloads and boom, done.

Usability to see changes live is a bit weird.

Ctrl+S brings up the browser save as but also saves the document and recompiles. Ctrl+Enter puts in a new line and recompiles.

Can click compile of course, but that's not efficient.

It's pretty simple.

You treat the AI like a junior.

You make it write pseudo code in the tickets.

Then you make it write code in the tickets for the tricky stuff.

Have it reference relevant documentation/APIs.

Then there is no lost knowledge.

(Still, sometimes it sneaks in some "helpful" belt and suspenders, but for the most part, I know everything that is going on in the code base with this method.)

And here I thought I was old.

So then GenAI seems like a lot of decisions on how to apply transformations.

And if I use an effect in photoshop, do I own the resulting image?

Still not understanding how this reaches limits. Do you have an example or analogy you can share?

What are you guys doing to burn through limits?

I have some dev + prod bots and according to ccusage, use the equivalent of $2500/month with them on CC yet I never hit the rate limits.

I feel like I'm using them all the time so I'm curious what you are actually doing that's burning all of these tokens.

Can you give me an example?

For me, it's:

  1. Write a spec for <feature>
  2. Add design for issue
  3. Write code
  4. Deploy code and manage configuration
  5. Run analytics