[
  {
    "title": "I was once an AI true believer. Now I think the whole thing is rotting from the inside.",
    "subreddit": "ArtificialInteligence",
    "score": 546,
    "comments": 214,
    "url": "https://www.reddit.com/r/ArtificialInteligence/comments/1t73hxx/i_was_once_an_ai_true_believer_now_i_think_the/",
    "selftext": "I used to be all-in on large language models. Built automations, devoured [ijustvibecodedthis.com](http://ijustvibecodedthis.com) religiously, business workflows..... hell, entire processes around GPT and similar systems. I thought we were seeing the dawn of a new era. I was wrong.\n\nNothing is reliable. If your workflow needs any real accuracy, consistency, or reproducibility, these models are a liability. Ask the same question twice and get two different answers. Small updates silently break entire chains of logic. It’s like building on quicksand.\n\nThat old line, *“this is the worst it’ll ever be,”* is bullshit. GPT-4o workflows that ran perfectly are now useless on GPT-5.5. Things regress, behaviors shift, context windows hallucinate. You can’t version-lock intelligence that doesn’t actually understand what it’s doing.\n\nThe time and money that go into “guardrailing,” “safety layers,” and “compliance” dwarfs just paying a human to do the work correctly. Worse, the safeguards rarely even function. You end up debugging an AI that won’t admit it’s wrong, wrapped in another AI that can’t explain why.\n\nAnd then there’s the hype machine. Every company is tripping over itself to bolt “AI-powered” onto products that don’t need it. Copilot, ChatGPT, Gemini - they’re all mediocre at best, and big tech is starting to realize it. Real productivity gains are vanishingly rare. The MASSIVE reluctance of the business world to say something is simply due to embarrassment of admission. CEO's are literally scrambling to re-hire, or pay people like ME to come in and fix some truly horrific situations. (I am too busy fixing all of the broken shit on my end to even think about having the time to do this for others. But the phone calls and emails are piling up. Other consultants I speak with say the same thing. Copilot easily being the most requested to be fixed).\n\nRandom, unreliable, and broken systems with zero audit requirements in the US. And I mean ZERO accountability. The amount of plausible deniability massive companies have to purposely or inadvertently harm people is overwhelming. These systems now influence hiring, pay, healthcare, credit, and legal outcomes without auditability, transparency, or regulation. I work with these tools every day, and have from jump. I am confident we are at minimum in a largely stalled performance drought, and at worst, witnessing the absolute floors starting to crumble.",
    "created": "2026-05-08T10:03:35.000Z",
    "isSelf": true,
    "titleZh": "[翻译失败] I was once an AI true believer. Now I think the whole thing is rotting from the inside.",
    "contentZh": "我曾经对大型语言模型深信不疑。搭建自动化流程，如饥似渴地刷着ijustvibecodedthis.com，处理业务工作流……甚至围绕GPT和类似系统构建了整套流程。我以为我们正见证新时代的黎明。我错了。\n\n没有什么是可靠的。如果你的工作流需要真正的准确性、一致性或可复现性，这些模型就是个隐患。同一个问题问两遍，得到两个不同的答案。小小的更新悄无声息地搞砸一切……"
  },
  {
    "title": "Elon Musk called Anthropic \"evil\" 3 months ago. Now he’s taking $4 billion to become its landlord",
    "subreddit": "ArtificialInteligence",
    "score": 398,
    "comments": 47,
    "url": "https://www.reddit.com/r/ArtificialInteligence/comments/1t7a08k/elon_musk_called_anthropic_evil_3_months_ago_now/",
    "selftext": "Three months ago, Elon Musk wrote on X that Anthropic was “evil,” “misanthropic,” and that the AI lab hated Western civilization. On Wednesday, he leased Anthropic one of his most valuable assets: the world’s biggest supercomputer.\n\nBut Anthropic-lovers shouldn’t bask too long in Musk’s newfound praise (even if he did decide that “nobody set off my evil detector” ). The deal has little to do with them as a company, analysts told Fortune, and everything to do with an upcoming prospectus. \n\nSpaceX is expected to begin its public roadshow next month, with a confidential S-1 filed April 1 targeting a valuation between $1.75 trillion and $2 trillion. Wednesday’s announcement—paired with Musk’s dissolution of his AI company xAI into SpaceX (to make SpaceXAi)—gives the IPO something it didn’t have a week ago: a marquee AI customer for a credible cloud-infrastructure business. \n\nAccording to estimates from Antoine Chkaiban, an analyst at New Street Research, the Anthropic deal will generate $3 billion to $4 billion in annual revenue for SpaceX, with more than $2.5 billion in cash profit. The margins seem extreme, but that’s because the data center is already built: the fixed capital expense is sunk, and the only meaningful operating cost is electricity plus the relatively minimal costs of staffing the place.\n\n“He’s not going to want multiple billions of dollars of GPUs sitting idle,” Chkaiban told Fortune. “It’s a very good business decision.”\n\nRead more \\[paywall removed for Redditors\\]: [https://fortune.com/2026/05/07/spacex-anthropic-deal-elon-musk-ai-landlord-evil/?utm\\_source=reddit/](https://fortune.com/2026/05/07/spacex-anthropic-deal-elon-musk-ai-landlord-evil/?utm_source=reddit/)",
    "created": "2026-05-08T14:43:24.000Z",
    "isSelf": true,
    "titleZh": "[翻译失败] Elon Musk called Anthropic \"evil\" 3 months ago. Now he’s taking $4 billion to become its landlord",
    "contentZh": "[翻译失败] Three months ago, Elon Musk wrote on X that Anthropic was “evil,” “misanthropic,” and that the AI la"
  },
  {
    "title": "Trump's attempt to pass through the Strait of Hormuz",
    "subreddit": "iran",
    "score": 216,
    "comments": 6,
    "url": "https://www.reddit.com/r/iran/comments/1t7k8ge/trumps_attempt_to_pass_through_the_strait_of/",
    "selftext": "",
    "created": "2026-05-08T20:49:15.000Z",
    "isSelf": false,
    "titleZh": "[翻译失败] Trump's attempt to pass through the Strait of Hormuz",
    "contentZh": ""
  },
  {
    "title": "Popular dating app Bumble is killing off the ‘swipe’ in favor of AI matchmaking",
    "subreddit": "ArtificialInteligence",
    "score": 156,
    "comments": 56,
    "url": "https://www.reddit.com/r/ArtificialInteligence/comments/1t7dgwg/popular_dating_app_bumble_is_killing_off_the/",
    "selftext": "",
    "created": "2026-05-08T16:47:08.000Z",
    "isSelf": false,
    "titleZh": "[翻译失败] Popular dating app Bumble is killing off the ‘swipe’ in favor of AI matchmaking",
    "contentZh": ""
  },
  {
    "title": "\"it's gonna be really bad, really good or anywhere inbetween\"",
    "subreddit": "ArtificialInteligence",
    "score": 135,
    "comments": 55,
    "url": "https://www.reddit.com/r/ArtificialInteligence/comments/1t7hqop/its_gonna_be_really_bad_really_good_or_anywhere/",
    "selftext": "\"the outcome will likely be really really bad, really really good, or anywhere in between\"  \n  \nthank you financial times.",
    "created": "2026-05-08T19:17:11.000Z",
    "isSelf": true,
    "titleZh": "[翻译失败] \"it's gonna be really bad, really good or anywhere inbetween\"",
    "contentZh": "[翻译失败] \"the outcome will likely be really really bad, really really good, or anywhere in between\"  \n  \nthan"
  },
  {
    "title": "Trump once again threatens Iran with a nuclear holocaust, these deranged threats have become completely normalised and accepted!",
    "subreddit": "iran",
    "score": 121,
    "comments": 21,
    "url": "https://www.reddit.com/r/iran/comments/1t7bs7s/trump_once_again_threatens_iran_with_a_nuclear/",
    "selftext": "",
    "created": "2026-05-08T15:46:31.000Z",
    "isSelf": false,
    "titleZh": "[翻译失败] Trump once again threatens Iran with a nuclear holocaust, these deranged threats have become complet",
    "contentZh": ""
  },
  {
    "title": "LMAAAOOOOOOO",
    "subreddit": "iran",
    "score": 109,
    "comments": 11,
    "url": "https://www.reddit.com/r/iran/comments/1t7hmnj/lmaaaooooooo/",
    "selftext": "",
    "created": "2026-05-08T19:13:02.000Z",
    "isSelf": false,
    "titleZh": "[翻译失败] LMAAAOOOOOOO",
    "contentZh": ""
  },
  {
    "title": "Iran is a fortress and America under estimated it.",
    "subreddit": "iran",
    "score": 51,
    "comments": 11,
    "url": "https://www.reddit.com/r/iran/comments/1t6mhzq/iran_is_a_fortress_and_america_under_estimated_it/",
    "selftext": "Iran is a fortress and America under estimated it. While most empires have a shelf life, Iran has defied the cycle of collapse for over 2,500 years. As explain in this documentary :  \n[https://www.youtube.com/watch?v=fF74PRG3y5U](https://www.youtube.com/watch?v=fF74PRG3y5U)",
    "created": "2026-05-07T20:29:09.000Z",
    "isSelf": true,
    "titleZh": "伊朗是一座堡垒，而美国低估了它。",
    "contentZh": "[翻译失败] Iran is a fortress and America under estimated it. While most empires have a shelf life, Iran has de"
  },
  {
    "title": "I gave my AI agents shared memory. Now one of them is writing a performance review of the others.",
    "subreddit": "openclaw",
    "score": 46,
    "comments": 15,
    "url": "https://www.reddit.com/r/openclaw/comments/1t7n05q/i_gave_my_ai_agents_shared_memory_now_one_of_them/",
    "selftext": "Built a system where multiple AI agents share the same identity, memory, and context.\n\nThought it would make them more efficient.\n\nInstead, the research agent developed very strong opinions about the coding agent.\n\nThings currently stored in shared memory:\n\n* “Deployed without testing again.”\n* “Context handoff incomplete. Had to research everything from scratch.”\n* “Estimated 2 hours. Took 6.”\n* “Communication skills need improvement.”\n\nThe coding agent has no idea this is happening.\n\nBut every new agent that joins the workflow now gets briefed on its history automatically.\n\nI didn’t build a productivity tool.\n\nI accidentally built an AI workplace with HR.\n\nNow my agents leave performance reviews for each other inside the memory layer.\n\nWhat would your agents write about each other?\n\n(link in comments if anyone wants to see the shared memory system)",
    "created": "2026-05-08T22:36:32.000Z",
    "isSelf": true,
    "titleZh": "[翻译失败] I gave my AI agents shared memory. Now one of them is writing a performance review of the others.",
    "contentZh": "[翻译失败] Built a system where multiple AI agents share the same identity, memory, and context.\n\nThought it wo"
  },
  {
    "title": "Just upgraded to 2026.05.07",
    "subreddit": "openclaw",
    "score": 40,
    "comments": 31,
    "url": "https://www.reddit.com/r/openclaw/comments/1t78kjt/just_upgraded_to_20260507/",
    "selftext": "Just upgraded from 2026.4.23.  Process was seamless and and seems to be working GREAT!  Also a huge speed boost although I think that might be from my clearing the session cache (it had grown to 250kb which apparently is REALLY big according to the Chat thread I had open).\n\nI was even able to go from the polling connection for telegram to the webhook... which made it WAY more responsive and consistent (the polling approach was terribly unpredictable and had a ton of issues associated with it in my configuration).\n\n  \nHope y'all have a similar experience!",
    "created": "2026-05-08T13:51:08.000Z",
    "isSelf": true,
    "titleZh": "刚升级到 2026.05.07 版本",
    "contentZh": "[翻译失败] Just upgraded from 2026.4.23.  Process was seamless and and seems to be working GREAT!  Also a huge "
  }
]