Maine Hacker News ke saath complicated relationship rakhi hai. Site geek news ka sabse important aggregator hai aur is blog ko traffic bhejne ka major source hai. Saath hi, is par toh toxic commenters ka fair number hai, jo meri direction mein insult uthata hai; agar aap taste lena chahte hain, to is article ko "watered-down" aur "slop" kehte hain.
Site geekon ke liye geekon chala jata hai, isliye tech trends se immune nahi hai; example ke liye, 2018 ke around, is par cryptocurrencies aur NFT par focus kiya gaya tha. Kehne ka khaana, recent shift feel karna zyada profound lagta hai: har din lagta hai lineup AI-focused stories, AI likhi gayi stories, ya AI par commentari dominate karti hai. HN AI singularity (21 July 2026).
Wo image khas bad din dikhati hai, isliye honest assessment ke liye maine February 2026 aur usi saal June mein systematic survey kiya tha. Feed ko AI-related topics se kaise occupy karta hai, samajhne ke liye maine poora February month ke liye daily top #5 ki sampling ki thi: AI ne 4 February 4 aur 12 February ko panch mein se char baatayein poya thi, aur alag se poori lineup 5 February ko (story #3 AI vendor ke liye submarine marketing tha). Sirf alag alag din jahan LLM news top 5 mein nahi the, wo the 1 February (pehla AI story 7th number par, phir 9th), 9 February (pehla 8th par) aur 25 February (AI 6, 9, 10 number par).
Dusri hissa experiment ke liye — jismein pata karna tha ki kounsi stories probably AI likh kar di gayi hain — maine Pangram use kiya. Pangram LLM-generated text detect karne ke liye bahut acchi, conservative model hai. Ye detectors techies ke beech badnaam hain, lekin objections often outdated assumptions ya outright misconceptions par based hain.
Tools ke liye AI writing ko koi bhi tarah "inhuman" ki zaroorat nahi hoti. Itne ke default voice current crop of LLMs quasi-deterministic hai: same essay do baar maangne par aap stylistically similar result paayenge. Individual mannerisms human-like hain, lekin bahut unlikley ki aapki writing usi set ki exact same combine kare.
Is baare main main thoda likh chuka hoon yahan. Results validate karne ke liye maine bhi koi flagged stories review ki thi aur maine socha findings sense banti hain; agar koi baat ho, toh Pangram bhi couple of false negatives rahi thi. Ye samjhane ke liye ki kya tha flagged, February 19 ke #3 story ko dekho (“AI not a coworker, it’s an exoskeleton”).
Isse 500+ upvotes aur 500+ comments mile. Meri opinion mein, ismein wide range of red flags hain. June mein, zyada details capture karne ke liye maine pure-play AI navel-gazing (vendor announcements, op-eds benefits ya drawbacks of technology, etc.) ko solid black diya aur hatched shapes stories jo heavily AI mein lean karte hain, lekin broader ramifications hain (jaise Instagram AI support agent account hack).
Jo stories sirf tangentially AI related hain (jaise RAM price hike reports) ko flagged nahi kiya jata. Pehle half of month mein roughly 60% of daily HN lineup AI-related ya AI-generated thi, jab tak hum month ke end ke paas gaye to ~50% thi. Isse upar jaata hai 40% se February se.
Is post par discussion Ready for more?.
स्रोत: Hacker News · HeadlinesBriefing द्वारा सारांशित