polytely

I think last time I looked it was that jet fuel is exempt from taxes in the EU? but trains aren't so flights are artificially cheap

yeah absolutely unusable, kinda insane to have access to fronteer models and produce a website that unusable

Yeah I feel like cheap enough AI will make having internet connected appliances such a huge liability that most devices will have to go back to as dumb as possible

yeah and I hope it accelerates the move away from using microsoft

it's very interesting how devoid celebrity culture is of videogame references, it is now finally slowly shifting, but it still is surprising when you hear a celeb talking about games.

my favorite one in recent memory is discovering that singer The Weeknd is really into indie games. (play Arctic Eggs y'all)

https://www.washingtonpost.com/entertainment/video-games/202...

The first time I realised gaming was actually happening everywhere and it was actually just that nobody was talking about it was when lady gaga was posting about playing Bayonetta.

https://x.com/ladygaga/status/1058894361486753792

Can you say anything about how much variance is there in energy use between models and effort levels. I'm very curious about the difference between the fronteer and things like GLM 5.3 flash.

its like a fox starting a chicken coop business

I see a lot of AI-pilled commenters in the YouTube comments saying it is just sour grapes, but I think there really is a difference between 2 human mathematician's scooping each other and a human + human + ai team being scooped by a full swarm.

I really feel like this is a precursor to these companies throwing their weight around in other fields. Like what if you are a pharma company spending years and years on a new medicine and then open ai gets wind and spends their considerable resources to scoop you. or your software gets cloned by them right before release. it feels like an anti-competitveness problem and I think it's probably best if a regulator steps in and forbids them from competing with companies/people in other fields? they should just rent out intelligence and not use it themselves for achieving stuff in different fields.

at least with GLM the banks can also use it to defend themselves for cheap, in the world Anthropic and co want everyone in the world is paying them an enormous sum in protection money every month to be allowed to defend themselves from hackers. it's such a racket.

yeah i just think that if I gave the initial pitch to Fable it would have successfully built something that wasn't exactly what I wanted. maybe it is because i went to school for design, but how something works is so important to me and it is hard to get to that point without actually thinking about it step by step and visualizing how interactions would work. I just cant imagine a single one-shot prompt containing enough information to for the model to know what to build. I guess if you have Frontier model money like these silicon valley freaks you could just iterate by asking it to tweak what you didn't like until it is right.

I'm mostly using deepseek 4.1 flash via openrouter, the way I'm doing stuff is:

1. write a sketch of a spec by hand

2. have the llm review the document and question me until it can generate a spec

3. review the spec and revise where needed

4. have it write an implementation plan

5. another round or revision/review

6. executing the plan step by step through the plan, plausing between each step to see if we are still on course and if the decisions it made track with my understanding of what we are doing.

I've been working for a couple of hours tonight, the total cost of the session is €0.6.

it's not the build this thing end to end, but also not quite write function x for me. It is still a lot of manual review, but I find I really need it to even discover what I actually want to build. I just cannot imagine building something in a single shot and getting something that actually has value (unless it is basically a clone of an existing thing). To me the whole value of ai right now is that it's now very cheap to build custom software that exactly matches your preferences.

It is because the people leading these labs don't want centaurs, they want to replace the human worker.

one of my projects on the backlog is to figure out how to combine the dvd copies and the bluray copies of the x-files so that I can have 4:3 episode in hd. especially in the first couple of seasons they crop a lot of stuff to fit it into widescreen, I'm waiting until AI gets good enough that I can using the dvd quality original fill in and upscale the cropped part of the hd episodes.

what do you use and why do you prefer it over omp

I love Elektron gear, I have a Syntakt and it is the piece of gear that I'm the most productive with.

Nice, Tim Schafer is someone whose writing I really admire, if you haven't checked it out yet Double Fine (the studio he leads) documented the entire (7 year!) development cycle of Psychonauts 2 in a 32 episode documentary series, and it in my opinion one of the best documentaries ever made. It covers a lot of cool stuff, like Covid, them getting acquired by Microsoft, big project leadership struggles. It's long as hell but it is really worth your time if you have an interest in creative work in large teams.

https://youtube.com/playlist?list=PLIhLvue17Sd70y34zh2erWWpM...

They are currently doing a company wide gamejam/game idea incubation thing, after surviving Microsoft undoing the acquisition and the same Documentary crew is doing a doc about that (they did a kickstarter campaign to fund it)

I sort of agree, and i think strengtening IP Law is probably not great. But I do think it's very fucked that building generative ai is only possible by taking the works of countless artists and craftspeople and then the model produced from that data immediately gets deployed to destroy the careers of the people whose, work was vital to it being created, without compensation for them, while making a few evil nerds richer than god. I think if you work at one of these labs you owe an enormous debt to society and your earnings should be redistributed among it.

how does this work from a liability standpoint? seems very risky for the person who is legally responsible for the business

Do you know the anecdote about Feynman and John Tukey doing a mental counting task and the one being able to talk while he's doing it and the other not, and then the reveal is they are using a completely different mental representation:

https://calteches.library.caltech.edu/3591/1/Feynman.pdf

worth a read, to me it kinda indicates that different abilities can be enabled by differences in how to mentally represent something. but yeah its debatable how much of difference in ability comes from/with a difference in experience but I my gut feeling is a lot more than you'd think. feels really hard to prove it as we only know what it's like to be us.

Also an Aphantasiac, the quote

> “we experience these as an awareness of a set of spatial facts.”

really resonates with me, as to how it feels inside my brain. I'm also really good at shape rotation and spatial reasoning despite not being able to literally picture the environment.

my pet theory is that there are just different places in the stream of consciousness your attention can rest and mine cannot be pointed at the 'mental machinery' that actually produces the images but instead rests at a higher level of abstraction after the visual signal is integrated into the representation of the 3d world that gets used for tasks like shape rotation.

I also don't have an internal monologue most of the time, but I can consciously will myself to do it, and then it is hard to get out of that mode, analogous to the 'you are now breathing manually' meme. I can do it but it feels annoying and slows down my reading and thinking.

Now that we have agentic systems the analogy of the internal monologue being a subagent is probably a good way to describe it, yeah i could read the thinking trace word for word, but most of the time it feels like existing at the level where the data that gets extracted from the stream of thought gets integrated.

Honesty it's so interesting to me how many different ways of experiencing the use of your brain we exist. Sort of reminds me of how there is no single method of crawling a baby learns, there are a bunch of different ways to do it and children just discover one and go with it. Walking seems to be different in that there basically is one good way to do it. I guess there are a lot of degrees of freedom in how the process of being can run on the brain.

which would be better

yeah their irresponsible behaviour doesn't match their stated concern about doom at all.

Remember that Chinese CEO that was executed for corruption & selling tainted baby milk powder [1]

I think these guys would improve their behavior if their actual life was on the line instead of that only being true in their less-wrong thought experiments.

1: https://en.wikipedia.org/wiki/Zheng_Xiaoyu

why don't you unionize with the other employees that feel this way across the mayor labs and pace the companies through labor power?

Don't you think open weight Chinese models will be the first thing they start banning once the regulation of this industry starts to kick in?

I've just been using openrouter to be able to quickly switch models and keep tight control over how much i spend and for that it's great. i also love their chat thing on the site where you can ask multiple models simultaneously , very good for brainstorming

the industry only exists because of cheating Americans, maybe fix your own shit first before passing judgement

Yeah if was in the US i would be very careful about talking to a us based llm about reproductive health or immigration stuff. no way they arent storing stuff about you that the government can easily get to

I'm guessing if you work on these images the whole day you start to lose the ability to judge image quality. like your brain becomes slopified

> So then I thought I would just photograph each page and have Claude OCR it. Tedious, but doable. Just to test it out, I went to the Claude web interface and asked it to do the first page. It refused, citing copyright law. Buddy, don’t you remember how you, yourself, were trained? Anyway, it suggested instead that I use a book-scanning service. Why are you giving me instructions about how to “infringe copyright” after refusing to do it yourself? (It didn’t believe me when I told it, honestly, that I had the rights).

not looking forward to the future where the things you can do on the computer are constrained by fear of breaking copyright in this way. The more AI reviewing stuff is put into processes we use the more annoying this will become. These AI labs are just really out there sucking up all our shared culture and then pulling up the ladder behind them.