It's one thing to write publicly and assume it'll be read by bots, but:
> Everything I write. Everything everybody writes is now food for LLMs.
This is the part of all this AI bullishness I find so incredible and disconcerting. Millions of people are sharing everything they write, their journals, their idle search queries, their medical diagnoses, their idle thoughts, the things they say to their friends and loved ones, etc. with glorified US defense contractors, plugging their Obsidian and Google Drive and chat groups in to an inference server that processes all of it...
Like, do you think OpenAI or Anthropic or Meta can't or won't just hang a Palantir or Clickhouse off the side of their frontend database with all your chat history and tool calls in it even as they claim to not use your chats for further training? Who's going to be the first to feed chat history in to a tool like Flock's Investigate tool? Whose going to be swept up in the next NSA Prism? Will it be me for voicing this concern on the clearnet?
> Will it be me for voicing this concern on the clearnet?
No, because you're no threat to any such powers.
For an oppressive regime, the optimal number of critical voices against it is non-zero.
It needs to be enough that people get to vent their anger before it metastasizes into actual actions, but it needs to be too little to spark threatening action.
You're striking that balance perfectly, meaning that you do exactly what it would want for you. => You're good.
>The smart way to keep people passive and obedient is to strictly limit the spectrum of acceptable opinion, but allow very lively debate within that spectrum — even encourage the more critical and dissident views. That gives people the sense that there's free thinking going on, while all the time the presuppositions of the system are being reinforced by the limits put on the range of the debate.
What do you mean metastasizes into actual actions, multiple nations already are creating legislature on AI, LLMs and privacy. Posting a comment on HN is not venting anger or frustration lol, that was a fairly level headed complaint compared to what people say publically about Musk and et all.
I think you're overestimating Dario's fanciful empire a bit too much.
Ya. The FBI report to the court said that Microsoft showed the GDID visited the ngrok.com/signip page while using a VPN. I would have figured at that level the OS would not know domains but likely IP addresses. So it must be browser telemetry right?
The sad thing is that before Spotify bought the Echo Nest[1], they had hosted some of the coolest discovery demos for non-mainstream (in my case ambient/IDM) where you would feed it a youtube video URL and it would make a really compelling radio playlist based off it. i found so many artists i still listen to today by just sticking a video in there in the morning and coming back to the tab when something incredible popped up.
When Spotify bought TEN i considered moving my listening over, but the radio button we ended up with in Spotify and Youtube Music are huge disappointments in comparison, so corporatist and flattened to 1.5 dimensions, I always wondered how the magic was lost.
Bandcamp's feed (especially once you trick the UI in to showing you how to follow tags) is usually interesting to leave running but limited in its own way by the artist pool lacking mainstream tentpoles to jump off of.
It's not easy, though. Most of my day job is spent trying to get html interactives on an e-learning platform to work reliably with iOS's ridiculous nonstandard interaction rules around when media is allowed to play. It's worse than working with the 20 year old jsp+servlet system that serves the interactives and business logic. no other browser behaves like iOS safari and to debug and develop against it you need an ios and macos device sitting on your desk. Firefox and Firefox on Android are a breeze but a rounding-error in our usage metrics, even accounting for our development. Apple desparately hobbles the web platform to collect IAP taxes.
> with iOS's ridiculous nonstandard interaction rules around when media is allowed to play.
Are there any standard interaction rules on when media is allowed to play? I thought everyone implements it differently based on their own ideas of security and user engagement
My first personal computer was a Palm and my earliest programming experiences came with reading these docs, how fun. For a young person whose parents' custody arrangement led to almost every night of sleep under a different roof, a wifi enabled OS with a great applicaton library that was always with me was really powerful and ended up being quite impactful on the arc of my life.
By high school I was writing apps that followed this hig with a fold up keyboard, designing the ui and compiling code on board the device. PalmOS 4 and 5 could be tricked out to be a whole computer, capable of working offline for a week and also could get you up all night on IRC and ebooks. it's hard to imagine using my smartphone offline for a week now....
and most of the apps implemented this hig and were straightforward to use because it was the defeult builtin toolkit largely the same its entire life, progressively enhanced GUIs from 1bit 160x160 to full color 320x480 responsive design... nowadays I'm building a Material3 app in my evenings and i know some day google is gonna make a material4 so that my app looks scuffed up
If one could swipe through the center without inserting a space, it would be incredible instead of perhaps only great... There was a PalmOS 5 keyboard like this named myKbd(1) based on some IBM research(2) which was quite fast to use. the atomik layout was quite quick to use.
> Everything I write. Everything everybody writes is now food for LLMs.
This is the part of all this AI bullishness I find so incredible and disconcerting. Millions of people are sharing everything they write, their journals, their idle search queries, their medical diagnoses, their idle thoughts, the things they say to their friends and loved ones, etc. with glorified US defense contractors, plugging their Obsidian and Google Drive and chat groups in to an inference server that processes all of it...
Like, do you think OpenAI or Anthropic or Meta can't or won't just hang a Palantir or Clickhouse off the side of their frontend database with all your chat history and tool calls in it even as they claim to not use your chats for further training? Who's going to be the first to feed chat history in to a tool like Flock's Investigate tool? Whose going to be swept up in the next NSA Prism? Will it be me for voicing this concern on the clearnet?