Let's assume I want every citizen to navigate the web freely while fighting propaganda machines as much as possible, so that means I want an automated system that creates the set difference between these two in real-time, as reliable as possible. To create such a system, and since there shouldn't be any overlap in these two sets, I can effectively put my efforts in half if I put my work in the detection of one such set.
The scaling problem, as I see it here, arises from the following: While the set of individual citizens (kind of) has an upper limit, represented by the number of internet users worldwide, the botnet nodes in propaganda machines do not. I can limit the set further, for example if I want to focus on the citizens that are part of my government only, whereas propaganda machines can come from anywhere on the globe. Internet users already need to provide a proof on authentication for quite a lot of services, while botnets generally want to avoid being identified as such.
While I'm far from in favor of Chat Control, I can somewhat understand why these initiatives are in motion at all.
> The EU is more of a threat to itself than Russia is
To put it mildly, this conclusion is non-sequitur at best.
It isn't even obvious to me which country GP refers to when they write "i already live in one". No reasonable individual would criticize "liberal politicians" and "electronic dictatorship" without making it absolutely clear where they are coming from. This obfuscation seems like a deliberate choice and makes any standing point balancing on crutches.
That seems like a horrible core idea. How is that different from data labeling or model evaluation?
Human beings want to help out other human beings, spread knowledge and might want to get recognition for it. Manually correcting (3 different) automation efforts seems like incredible monotone, unrewarding labour for a race to the bottom. Nobody should spend their time correcting AI models without compensation.
Speaking of evals the other day I found out that most of the people who contributed to Humanities Last Exam https://agi.safe.ai/ got paid >$2k each. So just adding to your point.
You and OP both work for the same "High Performance AI Inference" company, you might want to disclose that.
EDIT: and while you're at it, you might also want to work on your attitude. "you idiot", "get lost" and "you need to touch grass" are not helping any HN discussions
Well, in that case I'm curious... Why did you think hijacking OP's stance of "Claude did rewrote lots of my original messy code" with your own opposing position of "the project itself is not AI", and getting quite offensive about it, would benefit any discussion about this 13h old project?
It's a personal project of your dear ex-colleague, mind you!
> I come to Hacker News to avoid this kind of rhetoric.
I think this rhetoric fits seamlessly inline with the hacker ethos, and that's one of my motivators (if not the biggest) to read through HN comments at all. It's exhausting to comprehend at times, but so are any well expressed positions in the complexity of life. Otherwise, I worry that HN will complete its transformation to become just another marketing platform for the wider tech sector, like some seem to already think it is.
Sounds like GP is just in need for a G-Code to DXF converter when they mention "fringe stuff, cnc machine files from the 80's/90's" as answer to a sibling comment, though.
There are great FOSS CAD tools available nowadays (LibreCAD, FreeCAD, OpenSCAD etc.), especially for people who only need 2% of a feature set. But then again, I doubt that GP is really in need of a CAD software, or even writing one with the help of Gemini.
EDIT: or are you rather thinking about the book Working in Public: The Making and Maintenance of Open Source Software?
reply