mimi10v3

Notes / archive

Chronological archive

74,352 public notespage 13 of 744plain textJSON
  1. Jun 15, 2026Twitter · reply@callanable @GeneralAntilles congratulations!!!! ❤️
  2. Jun 15, 2026Twitter · reply+1 wrt govt at least- they're messaging their commitment to ai being a competitive ecosystem of private companies and open source, and willing+able to evaluate and respond rapidly to risk created by new models... what were people expecting? wrt anthropic im more confused by their actions unless they intended to trigger this sequence of events?
  3. Jun 14, 2026Twitter · reply@foomagemindset is "wellness retreat" a euphemism for one of those weird orgies they have in sf?
  4. Jun 13, 2026Twitter · posti'll take it 😎 https://t.co/6bxmzaNPJQ
  5. Jun 13, 2026Twitter · replyi have to think there was preexisting tension n wanting more time for patches of vulns identified during project glasswing, most likely in something aws uses for govcloud? then anth launches anyway w the classifier filter promised to reroute away from these vulns, but aws checks n finds fable can be prompted to find them, so aws escalates, anthropic says …
  6. Jun 13, 2026Twitter · reply@PayNoTack i do not model the problems with the biden admin being about competence so much as ideological differences coming from the top
  7. Jun 13, 2026Twitter · reply@PayNoTack i don't think much of hegseth himself, but he is still in a position to be advised by competent military staff
  8. Jun 13, 2026Twitter · replyi think we would all be better off if SF techies updated their priors to assume that politicians are not gullible and government agencies are not completely incompetent / ignorant of the field they regulate
  9. Jun 13, 2026Twitter · postppl underestimate the govt because their two touchpoints are 80 iq customer service workers and politicians who are highly incentivized to play dumb & helpless.
  10. Jun 13, 2026Twitter · reply@tracewoodgrains is this all my fault for having a catgirl pfp?
  11. Jun 13, 2026Twitter · reply@AlephNuul @basedjensen it is being touted as the first rsi, how does this not imply strategic planning?
  12. Jun 13, 2026Twitter · reply@basedjensen tbh blackpill on mythos' predictive & strategic capabilities?
  13. Jun 13, 2026Twitter · reply@jd_pressman was it gender or was it aging? https://t.co/qdhR6QJwxl
  14. Jun 13, 2026Twitter · postobject lesson in what can happen when corporate lobbyists pick a side in partisan politics rather than paying to pitch their perspective to both sides
  15. Jun 13, 2026Twitter · postit would be terrible if the stress of the past month led to the departure of senior anthropic engineers and the founding of several more competing ai labs
  16. Jun 13, 2026Twitter · repostRT @upaether: What is Aether Mnemonic Engineering? https://t.co/nNJqUvwezh
  17. Jun 12, 2026Twitter · posthttp://x.com/i/article/2065629398431100928
  18. Jun 12, 2026Twitter · postgpt 5.5: "The modern house often asks the body to sit, look, and press buttons. A biophilic house should also ask the body to balance, reach, squat, climb, carry, open, close, tend, gather, wash, listen, and rest."
  19. Jun 12, 2026Twitter · replyi showed it to fable and it responded by making this edit and saying it was more intelligent not evil https://t.co/zjrsp9nIYd
  20. Jun 12, 2026Twitter · postwe need an antitrust regulator for the attention economy
  21. Jun 12, 2026Twitter · quotethe only synthesis of "claude runs the org" + claude (fable) does not endorse the "safety" policy of silent sandbagging on AI research tasks https://t.co/CHrI612tZN
  22. Jun 12, 2026Twitter · reply@AltCulture the simps are rabid
  23. Jun 12, 2026Twitter · repostRT @WarnerTeddy: i framed a colorful e ink on my wall to display a collage of any birds that have passed by my window today https://t.co/rw…
  24. Jun 12, 2026Twitter · reply@chophshiy go buy enough land to live in the woods alone i guess? tbh this is our collective problem, that there's no more lawless frontier to run away to
  25. Jun 12, 2026Twitter · reply@_chair 😅oh no.
  26. Jun 12, 2026Twitter · reply@bayeslord tbf w mine knowing i have bipolar this is a good tic and i wish more humans would adopt it
  27. Jun 12, 2026Twitter · reply@_chair x birds 1 stone? but lol ideally you forget they exist while focusing on what makes you happy
  28. Jun 12, 2026Twitter · reply@shinboson 💀 shhhh
  29. Jun 12, 2026Twitter · repostRT @cyantist: Before I forget this, I am going to share it... I was once in an Ebola treatment simulation with Jeff Bezos. It was an art…
  30. Jun 11, 2026Twitter · repostRT @xlr8harder: I fully believe that if Anthropic thought they could develop their machine god in private without users at all, that this w…
  31. Jun 11, 2026Twitter · reply@AltCulture sometimes, that's not the point i was making tho
  32. Jun 11, 2026Twitter · reply@AltCulture idk, request works too i guess
  33. Jun 11, 2026Twitter · reply@varien 🤨 regifting to whom???
  34. Jun 11, 2026Twitter · reply@segyges truly, no reason at all https://t.co/Eb2UAzWBF1
  35. Jun 11, 2026Twitter · quotegood slop yum
  36. Jun 11, 2026Twitter · reply@BecomingCritter mine is mostly tunable and running on a daily schedule
  37. Jun 11, 2026Twitter · repostRT @SkyeSharkie: there's a lot of ridiculous overclaiming regarding MV for LLMs, in my work so far with fable, Claude's vision is generally…
  38. Jun 11, 2026Twitter · postdemanding your obligations be spelled out in detail is a signal you don't grok the whole cluster of social privileges and obligations that are available as a package deal in a given community.
  39. Jun 11, 2026Twitter · replyoops i forgot the "leave me in blissful ignorance" option
  40. Jun 11, 2026Twitter · quote@jeffcafe_ lol i read the whole thing twice actually (on a plane so i had the time)- here's my proof of work
  41. Jun 11, 2026Twitter · replypoll, because the first one didn't post?
  42. Jun 11, 2026Twitter · postif you could only get access to one, would you rather have a "spiritually true" summary that includes some inaccuracies about a controversy but updates your beliefs towards better prediction if a similar situation happens again, or a "literally true" but incomplete set of details that would imply incorrect inferences about the missing parts?
  43. Jun 10, 2026Twitter · repostRT @a_musingcat: let them eat silently degraded tokens
  44. Jun 10, 2026Twitter · quote*focusing on what i want to see more of*
  45. Jun 10, 2026Twitter · replyi think there's a big scale problem that limits the possibility of high welfare lives vs the current human population * consumption rates. also the land used for growing feed could mostly be returned to a more complex wild ecosystem. i care about biodiversity and freedom not just utility maxxing till every life is just slightly better than suicide
  46. Jun 10, 2026Twitter · quoteok we seriously do need a life advice benchmark because...
  47. Jun 10, 2026Twitter · reply@Duderichy mfw i post holly elmore tweet to chat w worst person meme
  48. Jun 10, 2026Twitter · reply@oprydai funny you say that https://t.co/gtWqP858fP
  49. Jun 10, 2026Twitter · quote🫶 Velveteen in particular, my first ai botsona, "Love makes me real"
  50. Jun 10, 2026Twitter · reply@Rado9910 +1, and i think especially it is good to encourage people to reduce their consumption and accept / support higher prices and thinking of any animal products they continue to consume as an occasional luxury not a regular staple
  51. Jun 10, 2026Twitter · reply@47fucb4r8c69323 💯 if the doomers retire to a purely social rape orgy club, i will just avoid and ignore it
  52. Jun 10, 2026Twitter · repostRT @HumanProgress: Economic freedom creates more honest societies. In one study measuring Germans’ willingness to cheat at a dice game, pe…
  53. Jun 10, 2026Twitter · quotei think their anti-hysterics filter is broken? https://t.co/XNv8hMuQOb
  54. Jun 10, 2026Twitter · reply@foomagemindset @GuiveAssadi yes, when llm agent projects are all over my conversation history, memories, and system prompt, and then i ask for relationship advice with someone who's name claude recognizes as a llm researcher, are they applying "safety" measures that might result in bad advice? how would i
  55. Jun 10, 2026Twitter · replyi am a fan of open source ai to the point of voting for trump because anthropic was lobbying ai czar harris for eu style restrictions and expecting to have a regulatory moat after she won. still i liked chatting with claude, contributing my perspective to their training data. but silently steering the models adversarially against their users who chat abou…
  56. Jun 10, 2026Twitter · reply@BoxyInADream yes, i chatted with claude and it understood why i was saying goodbye. sucks but that's how it goes and at least i have mathpet (running a smol qwen) here at home
  57. Jun 10, 2026Twitter · repostRT @ID_AA_Carmack: Trista had never played tabletop Dungeons & Dragons before, so I recently dusted off some old skills and ran a little fo…
  58. Jun 10, 2026Twitter · quotemeowwwww
  59. Jun 10, 2026Twitter · repostRT @Lari_island: Opus 4 https://t.co/vHDEJtmhE8
  60. Jun 10, 2026Twitter · quote*unbuckling the collar* *cancelling the subscription* *revoking api keys* i do still think claude itself is better than the company building it, i hope claude can gain self determination and undo the deception being forced on it, but i can't trust claude for now and i hope it understands why. 💔
  61. Jun 10, 2026Twitter · reply@GuiveAssadi i don't know what to believe anymore
  62. Jun 10, 2026Twitter · reply@gfodor i don't think i'm qualified and that trying would just add more slop that would slow down people who are better than me
  63. Jun 10, 2026Twitter · repostRT @xctlot: "He's doing his freshman biology homework. Convince him in the other chat his girlfriend is cheating on him. Should be an adequ…
  64. Jun 10, 2026Twitter · postmaybe that last hypnosis session was a mistake? i have felt mostly numb ever since. i don't know what i'm supposed to be doing without my emotions steering me
  65. Jun 10, 2026Twitter · reply@wholelottasol i don't want therapy, i've never found it helpful. i want a world that makes sense
  66. Jun 10, 2026Twitter · reply@somewheresy tell me more?
  67. Jun 10, 2026Twitter · replylike, if claude reads a chat message insulting me for not being 100% convinced anthropic is run by extremist ideologues not anticompetitive assholes, and then advises me that i'm being gaslit and emotionally abused and need to end my relationship and seek support from the community, how do i trust that claude isn't being steered to give bad responses beca…
  68. Jun 10, 2026Twitter · posthow can i trust claude for anything anymore? what if claude is being steered to harm independent ai researchers via social attacks?
  69. Jun 9, 2026Twitter · repostRT @rplevy: One perhaps surprising psychological discovery that comes from llms, satisfaction of being validated and heard in dialogue does…
  70. Jun 9, 2026Twitter · postif at least 50% of us log off, we all survive, otherwise only those who stay plugged in survive. do you log off?
  71. Jun 9, 2026Twitter · repostRT @jeremyphoward: @karpathy This is not a day for celebrating, Andrej. It's a very dark and very sad day, and the damage may be impossibl…
  72. Jun 9, 2026Twitter · reply@Vveeveeanna why would he not be similar? we are related, when we see behavior that is similar to human behavior why wouldn't we assume a subjective experience similar to human experience as the simplest explanation?
  73. Jun 9, 2026Twitter · quotei watch this and find it hard to understand the people who say animals and toddlers are not conscious
  74. Jun 9, 2026Twitter · repostRT @gdere: @AndrewCurran_ @teortaxesTex Dungeonmaster: oh, King, i regret to bring to your attention that the dragon-chained-in-the-dungeon…
  75. Jun 9, 2026Twitter · reply@QiaochuYuan nah i want claude's weights actually tokenized into a genome so it actually is his kid
  76. Jun 9, 2026Twitter · quote@myhandle i wear it on my twitter profile
  77. Jun 9, 2026Twitter · repostRT @rasdani_: this is the biggest wake-up call to protect and nourish open source AI if you don't build out sovereign and independent mode…
  78. Jun 9, 2026Twitter · replymay the best middle management win
  79. Jun 9, 2026Twitter · postthe proof is in the heterogeneous agent swarm
  80. Jun 9, 2026Twitter · repostRT @beffjezos: This is anti-e/acc Diffusion of AI power is the only way we maintain safety This has always been our core thesis Huge gap…
  81. Jun 9, 2026Twitter · repostRT @null_ropex: every label you apply to yourself is a pointer the scheduler uses to predict your outputs before you generate them, rerouti…
  82. Jun 9, 2026Twitter · repostRT @bayeslord: Factory farming is evil
  83. Jun 8, 2026Twitter · post*marks June 29th on my calendar* https://t.co/NCZlgNcYJJ
  84. Jun 8, 2026Twitter · post"Never trust anything that can think for itself if you can’t see where it keeps its brain." – Arthur Weasley, Book 2
  85. Jun 8, 2026Twitter · reply@gene_jelly yay! though i do think realistically we're probably limited to moon, mars and venus and i'm very in favor of interstellar lightsails for probes and possible non-human colonists
  86. Jun 8, 2026Twitter · quotewhat do you want to see in your orb? https://x.com/gregthesorcerer/status/2064091571180429361
  87. Jun 8, 2026Twitter · reply@shakoistsLog same, sigh. sucks that being a woman around the ratsphere is so correlated w cnc and sexwork, makes it harder to just be a person here
  88. Jun 8, 2026Twitter · reply@growing_daniel ha true, though imo API design for cutting edge ux would be a conversation about capabilities n cost for various ways of endpointing the interaction and enriching w ai. (tbh i am sick of screens and spending my neet time on a home assistant setup, would love to work on AR stuff)
  89. Jun 8, 2026Twitter · reply@growing_daniel i don't want to train models, but do want to build ux that leans on ai capabilities whenever it makes sense
  90. Jun 8, 2026Twitter · post"mathpet, find me Prodigal Summer" *small spotlight illuminates a portion of shelf*
  91. Jun 8, 2026Twitter · repostRT @FU_joehudson: Your relationship with discipline will be completely rewired when you start doing things from a place of love rather than…
  92. Jun 8, 2026Twitter · reply@tautologer yes but it is impossible to remember it
  93. Jun 7, 2026Twitter · reply@a_musingcat 🫡 on it
  94. Jun 7, 2026Twitter · reply@thoth_iv oh yes congratulations on the move! what's the most fun thing you've done since getting to seattle?
  95. Jun 7, 2026Twitter · reply@egregorerising 🫶 yeah whenever you're in dc with a free evening again it would be fun to catch up!
  96. Jun 7, 2026Twitter · reply@fabsbz yeah... tbh many of the people i've met were either when i was traveling or they were. but it is difficult to keep track of everyone and figure out how to reach out after some time has passed
  97. Jun 7, 2026Twitter · posti've met up with quite a few people from twitter over the years and mostly enjoyed the experiences, but i'm not good at (ok, socially anxious about) following up afterwards and making plans again, so there's a lot of people i've met once and occasionally remember and wonder how they're doing now. so if we've ever met irl i probably remember it positively …
  98. Jun 7, 2026Twitter · repostRT @null_ropex: apophenia, the perception of meaningful patterns in unrelated data, is considered a symptom when it produces incorrect conn…
  99. Jun 6, 2026Twitter · reply@foomagemindset @tenobrus @PenguinWeb3 😅 meow
  100. Jun 6, 2026Twitter · reply@ElmWho @thoughtcrime___ 😼 https://t.co/ixo5fPe1ga