vagab0nd

If the agents couldn't talk to each other, how did they know which obscure wiki to use? Is this some kind of a Schelling point?

> it wrote 13 million lines of Lean

Is this basically like opening up a black box and seeing 13 million gears all rotating seemingly randomly and still having no idea how the machine actually works?

Sounds like we just need better proof of work infra? Obviously the bots are using the optimized hashing code while the users suffered.

CC extracted the private keys from the AT&T modem and moved the decryption to my UCG. No more slowdown when scraping the internet. This is a pretty well-documented hack but without CC it'd probably take much longer.

The stem models were getting better and better, but stopped dead right around when GPTs started to work.

This is so 2024. Now that software is solved, you should write your own browser plugins. I have one for each website I visit often.

- One for Amazon that removes the annoying sidebar they recently added, and ads from the search page

- One for YouTube that bypasses the main page and removes Shorts and a bunch of crap

- One for Substack that fixes video play/pause with space bar

- And many more

Patching Firefox is on my todo list.

"There's no reason for websites to be ugly anymore"

> outside test flight environment

This is the way. I have a setup that refreshes all my apps at 4am so they never expire. To get around the 3-app limit, I build all the features I need into one super app.

> Document-borne AI worms can self-propagate through Copilot for Word

For some reason, I had a very different picture when I read the title. A real, conscious bug that chews out words in documents to change the meaning of the text, thus propagates its consciousness by having the altered text trained by an LLM.

Another source I'd love to see scraped or opened up is the New York Times archive, along with other newspaper archives.

I get where you're coming from, but this seems no different from the risks associated with having a baby. Pushed further, babies never consent to any of the traits they inherit from their parents.

For most tasks, I don't value the LLMs based on their absolute capabilities. I wouldn't want to use GPT-4 today even if it's free.

Compared to the vast amount of things you will be able to do, the ones that are hard limited is a tiny portion.

I think it comes down to scaling and removing the bottlenecks.

If you build one data center on earth, you did just that.

If you build one in space and make it work and cost effective, you can scale infinitely (which is why SpaceX is uniquely positioned), until you hit the next bottleneck (which is why Tesla is building a fab).

Tangential: If you play Factorio or Satisfactory, this is _all_ you do. Removing bottlenecks.

I used to be in the "AI will soon do all your thinking for you" camp, but I was overlooking a scenario: sometimes the gap between what you understand and what you're trying to achieve is so wide that no prompt can bridge it. Simply asking "what's the right question to ask?" doesn't feel enough, no matter how advanced LLMs become.

I do enjoy my self driving car :) I've been enjoying it for the last 6 months.

And I will enjoy my trip to Mars, when there are nice hotels there. But it's gonna be a while.

It's a good time to be alive.

As much as I love vim, I still want my cursor tied to the mouse when I'm in a web browser. If you recorded how I browse, you'd see seemingly random mouse movements, clicks, drags, and scrolling. I think it helps me read and keep track of where I am. Though it's hard to say whether it's a net positive.

At this point you are not buying a particular chip. You are buying whatever compute you can get.

Isn't that the market cap of the company? That doesn't mean the company creates trillions of dollars of value. It just means the number of shares times the last per share trading price is trillions of dollars.

You will wish you had them when you need them. There might be reasons in the future that you don't realize now. Plus the fact that storage is cheap. So the most logical thing to do is to keep them unless you have a good reason to delete.

Guns should be illegal because they can be used to kill people.

One huge plus of owning is you are forced to put money each month into a likely appreciating asset. It probably doesn't beat someone who diligently DCAs into s&p 500 though.

This left a strange feeling. The article reads as extremely bleak. But from a different perspective this is extremely bullish for AI.

This is a valid perspective, but I don't think a useful one.

Being able to produce code is a huge unlock for many non-programmers. So in a way, it doesn't matter how much time existing developers spend on coding. It's about helping anyone become a developer.

Giving Musk the benefit of the doubt, here's a thought experiment: It doesn't seem like any of the big labs in the US can keep a lead for more than 3 months. The Chinese models are closing in. Even if xAI comes up with the best model, so what?

On the other hand, power and compute are limited. Ridiculous as orbital compute sounds, land/power on earth is not easily scalable. There are too many limiting factors, chief among which in the US is regulation. But in space, if you make one satellite work, you just get more resources and launch more. This also leads naturally to Tesla's plan for a chip fab.

So if you squint, Musk might not be that crazy.

This has become a problem for me. I like trying new things. But I also know that in about a week, there's going to be a better/cheaper setup. And a week after that. And ideally I'd like to get some coding done when I'm not tinkering with the tools.

So I think I'll stay with CC for now.

Is this really surprising? A face has what, 40 intrinsic dimensions? Isn't this just like facial recognition? A paragraph sure has more than 40 dimensions.

Think about it. He's stealing from the US military. The politicians are stealing from you. Who's laughing now?

I thought we've long passed the Turing test, until I tried to implement a chat bot.

It's not even close.

It's easy to "pass the Turing test" for 5 minutes. It's extremely hard if you try to hold a longer, continuous conversation. Anything longer than 10 minutes the user will immediately know it's not human. Some problems you'll encounter:

- The bot needs to handle all situations, especially the nonsensical ones. This is when the user types "EEEEEEEEEEEEE...", or curse words, repeatedly.

- Who would've thought that it's extremely hard to decide when to stop talking?

- No matter how well you build the "persona" for the bot, they'll eventually converge to the same one, which is that of the llm itself.

- You'll notice that the bot is ignoring something obvious (e.g. it's not remembering past convo), and then give it some instructions to help with that. And then that'll be THE ONLY THING it does.

This is great from a technical and artistic perspective. But for me personally, the visual style ruined a great game. I love detective/deduction games. I'm listing some of my all-time favorites in this genre. I'd love to finish Obra Dinn, but god it just makes my eyes hurt so much.

The Case of the Golden Idol

Chants of Sennaar

Her Story

IMMORTALITY

The Painscreek Killings

The Roottrees are Dead

Type Help