8n4vidtmkvmk

Even so, I'm surprised it cost that much. I thought deepseek was cheaper.

But AI for installing tricky opensource software is indeed a good use case. I do that too.

What if you use 100GB of your 10TB plan? Is it virtually unlimited then?

It's all relative. Some people just can't use up their quotas with their normal usage.

If you're that worried, which you probably should be, download the docs into your project repo and don't give the agent Internet access.

How are you shipping less?

I believe the lower quality but is quality so bad you are afraid to ship it now?

I'm finding new things to demotivate me. It's good and quick at doing everything I ask but sometimes as soon as it comes together I realize I didn't really want this or it's just not fun. But maybe iterating quicker and finding that out sooner is still good. I don't know.

LLMs are good at cleaning up tech debt. Give them lots of small refactorings or dig out those crusty old P4 tickets. There's a lot of easy stuff for them that requires very little specification and very high probability they'll get it right the first time, especially if you can point them at an example done right.

You allowed to do that? You'd at least have to tip the box on its side if storing for a long time now? I'm no sommelier but I hear the cork needs to stay wet or they'll spoil.

I don't know what that's really trying to say, but it sounds wrong. That's how we end up with contextual interfaces that randomly shift around on a whim. Disappearing elements because some mysterious criteria has not been met. Atrocious for user discovery because the designer hid everything they thought the user didn't need.

You tried to discourage it from fixating on negatives by telling it not to fixate on negatives? And it didn't work?

Early versions of stable diffusion supported negative prompts and they never leaked because it would just down weight that stuff.

I don't understand why negative prompts never made it into the LLM world. If "no foo" and "dont do bat" don't work then just give me a separate textbox where i can put all my negatives!

It can be both biased and truthful. We've yet to see if it's false.

This example is not that niche. Lots of people use human to bash. I'd probably use a small pre trained model if it was easy to use. I use a little script right now that calls a cheap model. Actually.. $5 would probably will last me over a year so the only real benefit would be if I didn't have an internet connection.

Definitely need to optimize a bit for games and huge scale web apps. We've been finding big optimizations in our app recently. App works without them because we can scale horizontally but cutting CPU usage by 30% by eliminating redundant work and reducing copies of big objects? Why wouldn't we want to do that? This isn't even fancy algorithm stuff, mostly just shoddy initial implementations by 100s of eng working on a codebase over 7 years (not even that old). Stuff like that creeps in.

Why would they need help figuring that out? I can fully believe a decent LLM would figure this out on its own.

I had a flash model without vision capabilities take screenshots and convert them to ascii to "see" what was going on, all on its own. That's just one example. They're very determined.

Probably actually. I don't know anything about taxis but AFAIK its harder to get hired as a taxi driver so I would expect a little more training than Uber. Not a lot, but something

I didn't know people read him regularly. All i know about him is he made markdown and then cried when people tried to standardize it because it was underspecified.

Tbf, it's a little weird now with rolling restarts. I wouldn't call it downtime if there's a seamless transition to a new pod.

Maybe that's not what you're referring to, but do i mess that simple uptime metric... Finding out your server has been trucking along for 5 years without a single reboot.

I hope it remains this way then.

I am finding it still takes awhile to get my apps to a happy place with AI. Not because the AI is bad but because i have to sit with it for awhile. Figure out what's working, what isn't, what's missing, what turned out to be kind of useless and in the way.

As a different article said, we still need taste.

Yeah, but you're missing a gut intuition if something is off.

I wrote a fancy polygon decomposition algorithm in university (pre-AI) which my professor didn't seem very impressed by because it was missing some sort of mathematical rigor. Yet everything I threw at it worked! Even he couldn't find a counter example.

It took a while for me to find some failing cases but it turned out they did exist.

But hey, maybe all I was missing is an AI-written lean proof.

Are you defending doing boring work?

I personally never minded doing migrations. I guess I don't really have to anymore though. Weird.

Not from what I've seen. The desirable neighborhoods are 500-600k. Suburbs outside major city

For complex projects with lots of internal tools and strict requirements, I'm finding a fairly lengthy system prompt is quite worth it.

Start short or empty and watch where it makes mistakes then just keep tuning it so they're less frequent. That works for me.

In the US, permanent residents technically have to keep their green card on them at all times.

I signed up for AWS (SES) recently for the first time and find their whole thing confusing as heck too. Signed into the wrong region and they pretend they don't even know who I am.

Can you elaborate? Why wouldn't you just check in your code when you're done your bit instead of working on the same commit together?

I happened to "pair program" for the first time in years today! My coworker asked me to come over and help him with a problem.

'course we only did it for like 5 min to get him over the hump.

I think that's the only use case. It sucks having someone hover over you otherwise.

Amusingly i specialized in both AI and philosophy in college. Guess I'll be ok.

Just curious, why are drug names such a disaster? Do LLMs not understand pronunciation keys?

Better get the trophy ready

I too built a terminal emulator and I like it, but I wouldn't say it's easy to maintain even with LLMs. The # of edge cases is insane.