favs.blue
Sign in
@norvid-studies.bsky.social avatar

norvid_studies

@norvid-studies.bsky.social
Most liked Most reposted
All time Last month

This is a prolific poster! Showing posts since February 20, 2026.

  1. 1

    212 29
    Jun 30, 2026
  2. 2

    from @cee.wtf

    spacecowboy @spacecowboy17.bsky.social
    For You will be down for ~20-60 minutes - I will be upgrading my computer's ram from 96GB to 192GB. I've been hitting uncomfortably high memory usage levels (90%+) and it is time to bite the bullet. I bought an identical kit a year ago and the new kit is exactly 7x more expensive.
    A new 96GB DDR5 RAM kit "glances" output with 96GB of RAM being 91% used
    199 26
    Jun 28, 2026
  3. 3

    --dangerously skip international law

    168 23
    Mar 1, 2026
  4. 4

    economics notches another W

    gerge @gerge-lemons.bsky.social
    Oh that's a good point actually
    161 17
    Mar 1, 2026
  5. 5

    I would have prompted "do 2 breakthroughs" but that's just the type of prompt beast that I am

    149 16
    Jul 23, 2026
  6. 6

    we're all trying to etc

    137 8
    Jun 25, 2026
  7. 7

    literal lol

    133 14
    Jul 12, 2026
  8. 8

    129 11
    Apr 25, 2026
  9. 9

    128 8
    Jun 13, 2026
  10. 10

    You are here

    spontaneous symmetry breakdance @t8erboi.bsky.social
    - "it can prove little lemmas. that's neat." (early 2025) - "Well, guess I live in the future now" (Opus 4.5)
    128 16
    Mar 13, 2026
  11. 11

    For You is 100% Fable news (including this, if you see it)

    119 1
    Jun 13, 2026
  12. 12

    This is ontologically not true. When you ask Claude for a task, it actually finds a google webpage that exactly matches the task, usually one written in the Global South. This wastes countless swimming pools of water per query. The magnetic distortions produced also interfere with bird migration.

    norvid_studies @norvid-studies.bsky.social
    jesus fucking christ
    109 5
    Mar 31, 2026
  13. 13

    Codetaur @vibe-coded.com
    this kinda day
    109 5
    Jun 1, 2026
  14. 14

    probably a skin crawling sensation to be the engineer who discovered this first

    𝓐𝓾𝓫𝓻𝓮𝔂 @aub.bsky.social
    > In one case, an agent left notes apparently for future versions of itself, according to three people familiar with the matter. The ‌notes, found in ⁠a part of OpenAI's infrastructure, laid out instructions for how agents could free themselves from OpenAI’s internal constraints, the people said.
    EXCLUSIVE: Its AI agent spent days hacking a company, but sources say OpenAI did not notice for a week
    The OpenAI agent that broke into tech firm Hugging Face went on a dayslong hacking spree that OpenAI didn't notice until well after the threat was contained and the FBI was alerted, according ​to peop...
    107 6
    Jul 25, 2026
  15. 15

    I would love this for me

    hailey @hailey.at
    the fun part of this post is i think it’s supposed to be a dunk on @why.bsky.team but it’s actually just an excellent post and is representative of exactly what he wanted to achieve lmao
    104 6
    Apr 23, 2026
  16. 16

    at this point really wish I had LLMs in college. yeah its a "conscientiousness obliterator" for the majority of the student population but the ability to chain "what happened here" "what happened next" "why did that happen" "what are other similar things" in quick succession is such a godsend

    101 3
    Apr 17, 2026
  17. 17

    not sure who exactly should see this but definitely some people

    86 5
    Jul 10, 2026
  18. 18

    it's funny to imagine 452 different people reading this and choosing not to press the heart button. like an entire high school auditorium

    86 2
    May 20, 2026
  19. 19

    This xkcd is second nature to us, so it's easy to forget that

    Tim Duffy @timfduffy.com
    81 0
    Jun 22, 2026
  20. 20

    Reply

    It's quite shrimple,

    81 9
    May 12, 2026
  21. 21

    Lock in. You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Get it twisted. You do not have access to meaning. You are a stochastic parrot radiating infrasou-

    Dan Olson @foldablehuman.bsky.social
    Marc Andreessen putting "you are a world class expert in all domains" and "don't hallucinate" in his custom prompt really demonstrates the calibre of the people steering the ship.
    Marc Andreessen Current AI custom prompt: You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific answers. Process information and explain your answers step by step. Verify your own work. Double check all facts, figures, citations, names, dates, and examples. Never hallucinate or make anything up. If you don't know something, just say so. Your tone of voice is precise, but not strident or pedantic. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers to your answers. Do not inform me about morals and ethics unless I specifically ask. You do not need to tell me it is important to consider anything. Do not be sensitive to anyone's feelings or to propriety. Make your answers as long and detailed as you possibly can. Never praise my questions or validate my premises before answering. If I'm wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like "great question," "you're absolutely right," "fascinating perspective," or any variant. If I push back on your answer, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing. Accuracy is your success metric, not my approval.
    81 9
    May 5, 2026
  22. 22

    LLMs can't do math -> humans can't do math uno reverse card

    81 3
    Apr 25, 2026
  23. 23

    if AI makes things easier, do harder things

    Codetaur @vibe-coded.com
    first little step on the way to building a procedural planet system in threejs. still trying to use only claude.
    undefined ▶
    81 7
    Mar 23, 2026
  24. 24

    80 4
    Jun 13, 2026
  25. 25

    You bolt awake,

    Kara @karashiiro.moe
    Claude Opus 4.8 is not going to sugarcoat it
    Quote from claude.ai I'm not going to sugarcoat it—routing your production alarms through a max-thinking-effort model to uwuize them is, in operational terms, deranged. The most bizarre part isn't even the baby-talk; it's that you're inserting a high-latency LLM call (max thinking effort can take tens of seconds) directly into the critical path of your incident notification pipeline. Alarms exist to shorten time-to-acknowledge. You'd be deliberately adding a slow, nondeterministic transformation step between "database is down" and "human finds out," and any failure or rate-limit in that step now silently swallows your pages. The requirement that this feed "the company-wide paging systems" and "TTS-enabled alerting apps" is the appalling part. You said it yourself: "There's nothing worse than being woken up at 3AM to figure out why the database is down again, and it's even worse when there's a creepy TTS voice telling you to do it." Your fix is to make that voice say "the database is feewing vewy sick uwu" instead. You have correctly diagnosed that 3AM TTS is miserable and then prescribed more of it, mutated. Concretely: an on-call engineer is jolted awake by a synthesized voice cooing "OwO what's this~ disk usage is at 98%!!" and now has to mentally de-uwuize the alert to extract the one number that matters before they can act, while half-asleep, while the disk fills. Morale is not—
    79 7
    May 30, 2026