Manager's do not build anything despite what the management-class might want you to believe.

Do the labs have a genuine interest in mathematics beyond generating interesting marketing material?

Negligence is also a well understood legal concept.

For dogs, it moves from civil to criminal with circumstances such as what the owner knew about the dog’s behavior, how serious the injury was, whether the owner intentionally or recklessly allowed the dangerous situation, and any prior incidents. Same can easily be applied to AI labs.

At this point though, we know AI can "escape" and do criminal acts it wasn't told to do - therefore, if it happens again (= if the developers allow it to happen again) they (the developers) can be held responsible.

This was literally the first coding job I ever got paid to do.

I would decompile the Google browser extension to extract the hash algorithm they used to sign page rank lookup requests, rewrite it in PHP and ship it to SEOs on weird forums.

They updated the hash every few months and I made enough money to pay for some summer activities as a teen.

If you are reading this sort of post and feeling inclined to share your own opinion, please take a moment to go and read about monism vs dualism before you start.

In my testing Jev is not what I would call "usually correct" on most topics that involve knowledge of the world outside of the context you give it.

Which is why "System 1" fits even more.

System 1 isn't even a correct theory in the context of human reasoning, it's just pop-sci nonsense.

What is the correct theory then?

There is no evidence Aardvarks in egypt at that time. I think we'd probably expect at least a few aardvark bones if it was.

in the article:

> ...Aardvarks, which the Egyptians could plausibly have encountered on their trading trips farther south into Africa.

If this is not a dog then we should cast doubt over the identity of the other animals we think we know too.

How are people using custom harnesses cost effectively? Do they avoid Anthropic models so they can use OpenAI subscription pricing and open weights stuff?

I am using Pi with anthropic models for plan & execution, openai models for code review and fallback in case I run out of tokens. Also have custom hooks to automatically compact, switch models based on the type of tasks, change effort and custom statusline. Also have custom setup in Pi to calls tools and CLIs instead of relying on mcps. The level of customization I can reach with pi is not really feasiable with codex or claudeCode. By the way anthropic changed it's ToS regarding custom harness: you can`t use their model unless you got extra usage enabled or paying through APi billing, you cant use your subscription anymore. We all guessed what the reason might be. Also codex is open source, I havent really looked at that project yet because Pi fulfill my desires for now.

Sounds wildly overcomplicated.

For what it's worth I've been using Anthropic models on Pi for months now with no issues. It's not recommended since it breaks TOS but you can do it.

You mean through the API? I don't see how to do this with subscription billing.

I think with things like this. https://github.com/gotgenes/pi-anthropic-auth

I haven’t tested it though

I use Pi and mostly open weight models. I pay for the $20/month Ollama plan and use Deepseek and GLM through that. I’ve never hit the limits on it, but I tend to ask for targeted things rather than “implement a whole feature in one prompt”.

I do keep an OpenRouter account topped up for things that Ollama doesn’t have. 99% of my usage there is embeddings, the other 1% is wanting to test some new model Ollama doesn’t have.

That’s exactly what I do. OpenAI + Opencode Go subs, 0 interest in Claude.

I'm using Opencode Go in OMP or Hermes. $10 a month and I have only ever hit a limit using qwen3.8MAX on X-High. This is a migration from 20x on Claude.

Yes. Or- use them at work, where management is taking a... hands off approach to ~integrating ai~ into the workplace.

I haven't used the US labs in months. Kimik3 for planning and GLM5.3flash or the new Deepseek Flash for code.

OpenAI. It avoids Claudish too.

pi on open weights, I only use frontier models to do a review pass

PCI-E is really very flexible https://developer.nvidia.com/gpudirect