hahahaa

¿Por qué no ambos?

Bye bye event horizon.

why not just use irc?

Half joking, half serious. We need AI to challenge assumptions to not just "you are absolutely correct" all the time.

Yeah silly if ruled otherwise. It would mean copyright defaults to the laggiest country since someone could throw a paper plane with the material over a border.

Whataboutism moniker doesn't apply when the context is "avoid Chinese models because China commits crimes against humanity" and context is also "Top sota models for useful work are created by, let's list them all: US and China". If the argument is don't use Chinese models because China is hostile then let's say "The USA.... Objection: Relevance? Overruled".

US is less hostile in the day to day sense, but they could look at your cloud data at any time because say someone in your company protested and exercised their 1A right, and the government didn't like that, so that sort of thing tracks directly to secrecy and privacy guarantees I might want from models.

I personally don't worry for my sloperating but if I ran a large company's AI I might consider it. Esp if in Europe.

Slack isn't responsible for the work product of its customers. They provide a service, and like to get new customers. Those customers use agents. Which tech solution they use is debatable but they can't ignore it. They have supported bots and 3rd parties for eons anyway.

It is like asking in 1999 why DHL should have a website. Surely they can just meet people where they are - on the phone, on the high street. Why does everything need a bloody website.

Yeah just need to tell the system prompt to not do that. A grumpy senior engineer agent that pushes back on constant tech stack changes would be good too.

I want an LLM trained on dailywtf code

And how is US doing on digitally totalitarian and hostile towards its own people?

In practical terms you could get US and Chinese models to review each other, right. Depends what your use case is. Coding is kinda not so bad it is reviewable and immutable/traceable per commit. An AI app that is like a psychologist or something may be more worrying.

Try different harnesses people! I am actually preferring Chinese models at a fraction of the frontier price for coding. Yeah you need more tokens per unit of work done, but it is way cheaper still. Using CC/Opus as a staff eng / frac CTO. And Hermes/Chinese model as hopefully my team of mid levels. This way I can make good use of pro plan and then get cheap Chinese tokens for the rest and not hit a RL and know it can scale up. plus choosing your model is so cool and some are a lot less verbose.

Hermes is a better coding tool IMO. I can't put my finger on why but it just feels better. Maybe being true yolo helps.

Also incredible horizontal scale.

and vertical scale

and fuck fuck fuck load of features.

It ain't cheaper on AWS lesson getting relearned everywhere.

Dream of enshitification.

An easier Amazon Farmgate?

I wonder if those were fast enough nostalgia says yes but I bet it was dog slow if we tried today with 2026 lens.

Yeah I don't get it we had terminals on single core machines. In early 2000s I am sure they could handle the UI of Claude (probably IO too).

Is that after costs? Are there much costs? I guess repairs, insurance, rates, taxes, power etc.

Thanks. I am guessing you have to try stuff and build tacit experience. No other way, just get stuck in and try stuff, then try and learn bits from others?

There are other things to consider. Free public transit would make it more desirable to not use cars as much reducing congestion, accidents, pollution, energy usage. The need to build less high traffic roads may be a money saver to offset. Also it will get more people to exercise by walking to bus stops. You'd need disencentive for abuse but you can personally only add 16 person hours of usage per day so it is not like say free unlimited money or electricity where a single user can add a lot of strain. And spending all that time on transit is a cost. For homeless people yeah you may need to solve that problem! Build more housing for sale and rent driving down prices and reducing the number of homeless yet productive people (with jobs but can't afford rent). Build shelter for homeless with problems they can't hold down a job.

Right maybe we can make them free and safe? Not sure why that is not possible. Walking on the street is free. In many places it is safe too.

500k memory but not sure about disk.

Yep I noticed this when starting a reading group at work. High fomo, queue formed. And that was not even an AI one.

This is an easy move too. I.e. it is hard to have impact in the technical sphere when everyone is doing that.

Organising something (book a room etc.) is what engineers tend not to want to so so it becomes a form of arbitrage. Less mental effort and more visibility.

Will you consider franchising or expanding to other countries?

Not because of his net worth (he is probably worth 200bn if it came to liquidating) but because he will pull some trick to get the money. Maybe he will chuck SpaceX under debt, draw a dividend then set up a 1000 year payment plan with the bank.

The Federal Reserve? You ressurect Alan Greenspan, and tell him to brrrrr.

That's fine. Like they say when you owe $1m it is your problem but if you owe $100m and can't pay it is the bank's problem.

I do a pls sometimes.

Yes absolutely. A good analogy would be Gibhli. If you saw a Gibhli styled ad for say your local restaurant would you consider it art.

Now rewind, uninvent AI and then would you consider that same ad art?