jedberg

The whole paragraph above explains it. Radio through your head.

Terraform deskilled us too. No one is saying, "SREs should configure their infra by hand sometimes to not lose that skill".

If AI can handle it, great! We have a new tool -- let's make the tool better.

Let's figure out how to use AI to make us detect and resolve issues faster, and even better, prevent issues. I've always said we can't consistently hit 5 9's without automatic detection and remediation. We are there now.

Let's get everyone to 5 9s with AI!

No one in Google is lying. If what the person in the Twitter post says is true, then everyone at Google is telling the truth, and so is OP.

And given all the replies to that post on Twitter, it sounds like it's true.

You should submit this as its own story. This is big if true, so they say (and based on the replies, it is true).

Sure, you could operate without DD. You can also operate without advertising, but it makes it a lot harder to reach customers.

Elsewhere in this post a store owner talks about how DD is a necessary evil. How they have convinced a big chunk of customers to only use DD to order food, and you will lose out if you aren't on the platform.

> What I don't understand is why DD gets any cut of the store's profit. T

In theory it's because that customer wouldn't be purchasing without DD. In practice it's a shakedown.

No but at the current price you can just carry two phones. :)

That's why "They have orthogonal capabilities" matters. You need to find the one that's best suited to your particular task.

But the model names are linear by price.

Part of the confusion is that it's not linear capabilities. One model is not strictly "better" than the other. They have orthogonal capabilities.

I think the names are more of a reference to price. Luna is always cheaper than terra which is cheaper than Sol which is cheaper than Astra (for the same model number).

So you don't have to do the math, at a midline valuation of $1.7T, the Amodei's each will have about $25B. It mentions that they have pledged 80% of their stock to charity in a non-binding agreement.

This is interesting advice. When I was coming up as a sysadmin, the advice was always "run a home lab it's the best way to learn all the details". So I did and it was helpful.

But you never heard someone say "you should still code in assembly so you can understand what your compiler is doing".

Is this more like case one or case two? Will it matter if you understand in a few years?

Or maybe Codex/Claude Code got better and more people are just using those?

Or more people are using harnesses that do what openrouter does so they don't need it anymore?

Or they are using other GPU providers directly to run their open source models?

I don't think this data is conclusive. It's a very unique subset who are using openrouter.

Thought it was great, didn't want to pay. We made it free for the basic todo, paid for groups and templates.

We also didn't have the advantage of good whisper models so voice input wasn't a thing. I feel like that hurt adoption as well.

I've been a Principal Engineer for a while, so this is how I've worked for a while. Sometimes when you're a PE, you either design something and then junior engineers implement the details, or you're dropped into a project that was already designed and built by more junior engineers, and you have to quickly understand the design an as many details as you can.

With AI, everyone now gets to act like a PE. Your job is to design, the AI builds like the junior engineers. Then you review their work. You may never understand the details, and that's ok, as long as you know the main ideas and where to ask for the details (in this case the LLM instead of a human).

I tried to build a todo app a few years ago with a similar thesis -- that the problem with todo apps was triaging the list.

So we built an app that did it for you, just like yours, using machine learning and AI (and some simple rules based on due date and expected time to do the task).

Turns out people don't want to pay for that, even at $2/mo. We even had group features, so that a parent could put tasks on their kid's list, or a group of students could use it for a group project, or a team at work could use it to coordinate team tasks.

We also had templates -- tell us you're a single person who lives alone, and we would auto-add tasks like washing the sheets and cleaning the toilets. Or we could add a daily exercise routine or even a daily bible study if you told us you were a religious Christian.

Feel free to take these features. I hope you have better luck than I did.

I've been using it for a long while in SF. I love it. Never had a problem.

I know of one CEO who said "If you copy/paste an AI response and send it to someone else, they didn't need you in the loop and you don't need to work here anymore".

1. I didn't know it exists. Now that I do and I looked into it...

2. I have an M1 chip and I'm not sure it's supported.

3. ChatGPT can open both browers, not sure if Safari plugin builder can

But if it does all those things, great. My point was more that plugins won't be a problem anymore because AI can build them.

I use all three browsers on a daily basis. Chrome for accessing google owned properties and things I click on from Google search, Safari for reddit and hacker news, and Firefox for everything else.

Safari is always the best on my Mac, but my complaint has always been the lack of plugins.

But just now I ported one of my favorite plugins from Chrome to Safari using ChatGPT. I told it I already have it installed in Chrome so it can get the source code there, and then told it to port it and install it.

After 20 minutes and few bucks in tokens (estimated, I actually used my 6% of my $20/mo plan), I have Session Buddy in Safari.

So maybe the plugins aren't a problem anymore...

> Wonder if the law firms are saying the same thing about software

If they're not, they should be.

> Will you still say it's great when it is the society as a whole?

Yes. A world in which menial tasks are taken care of by computers is a wonderful idea.

The important part is that our society needs to transform around that. If all capital accrues to the owners of the AI, then we'll have a problem. If we remake society so that everyone shares in the success, that's a very different idea.

> Impossible to follow any topic without it devolving into political flame wars.

But have you considered how that is Donald Trump/Joe Biden's fault?

/s in case it isn't obvious.

Not good for the law firms, but great for the rest of us!

> but we really need to think about how the next group of people are going to learn the skills and nuance to applying them

That's what school is for. For the last 150 years we've automated a bunch of things, and then we send people to school to learn how those automations work, so they can make better automations.

Maybe law school will have to be a year longer. Or maybe some forward thinking firms will have new associates check the work of the LLM to gain that experience.

Way back in the day, everyone wrote assembly. Then things like C came around, which had to be compiled to assembly, and most people objected: "I don't want to be that far from what the CPU is doing!".

Each time we move up an abstraction layer, people object, saying that nothing will work right if we move up that high in the stack. Give it time. I'll bet in 5-10 years, almost no one will look at code anymore, except in special cases.

We still have people who write in assembly, we still have people who write in C, mostly for performance reasons. Because you still need to be close to the CPU to get that last bit of performance. But in most cases, it doesn't matter. It doesn't matter if the LLM makes bad code if it works and passes all the tests.

Granted, this is the first time we're going from deterministic to non-deterministic, but is it really? Large scale distributed systems are already non-deterministic. They create inputs and situations that programs aren't designed to handle all the time. What's a little more chaos? :)

I would love to see these benchmarks split up across "generic" tasks and novel tasks.

For example, redlining an OEM agreement with my vendor. They pretty much all use the same template, and the objections are always the same. I tried using Claude first, and even my lawyer was impressed that it found things they missed. Then I used Claude to redline that type of contract and the other side said "your lawyer was surprisingly thorough". I had my lawyer do the final check, but saved a bunch of money doing the first three rounds with Claude instead. And my lawyer basically just said "looks good to me" by the time they reviewed it.

Or even writing a simple will or estate document.

It seems like LLMs are great for the kind of generic work you'd usually pawn off to a new associate or paralegal, but not for the important things that partners usually do.

Yes, my spouse has access to my password storage, which has all the decryption keys. Whether she remembers how to do it is a different question. :). Maybe I should write it down, or show one of my kids...

Sadly I don’t have a family member that I can trust with that, but I do put an encrypted copy on AWS glacier and I have an encrypted volume in Dropbox.

I moved homes and damaged both drives because they traveled together.

> Self hosting my photo albums

I used to do this, until I learned the value of offsite backups.

Learn from my mistake (if you aren't already :) ).

Have you heard the phrase "reality has a left wing bias"? People perceive highly factual information as left wing.

For example, if I told you yesterday the President made a statement, and then repeated the second half of that statement, which is a sign of dementia, you'd probably accuse me of being far left, even though that information is just simply factual.

That plus the fact that if you dig into the reports, the two that said "barely left" even admit that they only chose that because their rating system doesn't allow for "center".