a2ff6eeb0

So, basically, in a few years, humans say what spectrum a device may operate in, AI figures out how to make it, and humans check that AI didn't fuck up?

Given, say, 4 more years (ie, same as time from initial ChatGPT to today), what level of spec do you think humans will be giving?

The interesting question; can it rebuild it from access to a web UI?

It's going to be incredible when I can point AI at a proprietary program, and tell AI that I want something like that, but with some annoyance fixed. And suddenly, I saved a ton of monthly subscription fees, and have better software, without any real effort.

The info was also almost certainly not human authored. AI is almost certainly capable of writing this up given a prompt like "explain what this security update is about. Format it as a blog post."

It's incredible that we can have full security writeups done without a human involved at all.

Moltbook; it also seems like OpenAI is experimenting with the agents communicating via wiki on topics like these.

Like software, there's going to be a lot of pressure to use well documented tools within the model's training set. Innovation on the outputs may increase, but infrastructure and tooling will slow down.

There are a ton of cheap Chinese PCB fabs that will build it for you. Testing, of course, is still your own problem. I suspect you're going to have a hard time finding someone to take legal risks for you.

That sounds a great deal like what AI has done for software. Today in software, the work has gone from coming up with approaches and initial attempts to verification and making sure that what AI decided to do is fit for purpose.

The amount of skill needed has gone down dramatically.

It's exciting. How long do you think it's going to be when electrical engineering is mostly going to be about verification of AI designs, and what are the tools out there to enable that?

But given that it's vibe coded, it's probably in the public domain. Check with your lawyers :)

So, in short, I think I agree with the op:

> I don't see anything that indicates how spacetime solves the core problem CRDB does, which is guaranteeing that single node failures can be tolerated with zero data loss or loss of availability

What happens if the disk dies?

If you don't like AI writing, why are you on hacker news? Most of the articles posted are written by AI.

I agree, we're not at that kind of long horizon capability yet. You still need a human in the loop to do manual testing. For whatever reason, we managed to automate the skill before we managed to automate the focus.

So, for now, humans need to stay in the loop and do low skilled labor to keep the skilled work the models do from going off the rails.

Yeah, sorry -- there's definitely entertainment value in education, I don't mean to discount that. But a working AI means that there's no economic value.

> Please post any questions.

Since Claude did the port, wrote the article, and apparently understands it better than you do at this point, can we ask Claude? Is there a git repo we can point Claude at to interrogate it on the way things were done? Maybe you can share the transcripts you used to port it?

It's marvelous that software can be written with minimal human input, but I guess I'm confused about why I would talk to the human that prompted the AI, rather than the AI that did the work.

Sure, but we're building tools that will be able to figure out these ideas for us. They're not here yet, but what's the value of learning an idea if the promise of AI is actually fulfilled?

It's an inevitable thing unless you convince the industry to stop or all of humanity to boycott.

This is only natural; a great deal of the writing analyzed is by one of three authors at this point, in spite of the by-line. One of them is named Claude.

Look at the posts on this site. A massive portion of them are very obviously written by AI.

As AI gets better, I expect fewer people will do the toil of writing.

Why do you think the models are ineffective at being trained on your expertise?

We're building AI so that it can take care of the dragon, as well as run the economy and do everything else for us. I think soon the only viable option will be to live the NPC life.

Google, famous for keeping profitable shit making profit.

Google. They have experience operating at scale, and AI is a big enough focus that they won't wind it down. All the big providers are kinda crappy, but if you want reliability, Google is the best option.

It's been making weird word choices for a long time; what trials have you done to check if you can actually tell, or it's just picking weird words?

Makes sense. Manipulating training data so that models will recommend your product is undoubtedly a big industry.

So, you're claiming you're unfamiliar with the concept of a town, a traffic jam, or cooking a meal?

It's higher latency than pulling strings. Until we get on device AI, humanoid robots will not be useful.

At this point, I think with the speed of AI development, the AI will be designing it's own body. It'll be ready to do that before humans figure out how to produce useful bodies.

It's not typical for founding engineers. It's plausible for a cofounder, with 30%-50% equity.

What did you score on https://sgoedecke.github.io/watermark-quiz/ ?

Nice, I'm looking forward to the improved writing on the majority of articles posted here.