[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-grokipedia-turns-ai-wiki-into-copy-loop-en":3,"article-related-grokipedia-turns-ai-wiki-into-copy-loop-en":30,"series-tools-620b495e-d447-4b5e-8477-f75248729685":83},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":22,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":29},"620b495e-d447-4b5e-8477-f75248729685","grokipedia-turns-ai-wiki-into-copy-loop-en","Grokipedia turns AI wiki into a copy loop","\u003Cp data-speakable=\"summary\">I break down Grokipedia’s launch, bias, and copy-from-Wikipedia workflow, then give you a reusable AI-encyclopedia template.\u003C\u002Fp>\u003Cp>I've been watching AI knowledge bases get built the same way for a while now: big promise, messy launch, and a lot of hand-waving about “quality.” Grokipedia is the latest version of that. On paper, it’s supposed to be an encyclopedia. In practice, it reads like a product team trying to out-argue Wikipedia while borrowing half its scaffolding. That’s the part that bugged me. I’m not even talking about the ideology first. I’m talking about the workflow. If your source of truth is a model that can rewrite, summarize, and “fact-check” itself, you’ve already invited drift. Then you add copy-forward from Wikipedia, a custom license, user suggestions routed through Grok, and a launch that had to be delayed for content quality issues. Yeah, that’s not a clean system. That’s a pile of tradeoffs pretending to be a knowledge product.\u003C\u002Fp>\u003Cp>The source that kicked this off for me was the \u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGrokipedia\">Wikipedia article on Grokipedia\u003C\u002Fa>. I’m using Wikipedia here because it’s the most complete public summary in one place, and it already stitches together the launch history, content model, criticism, and follow-on coverage. I’m also linking the underlying companies and tools where it matters: \u003Ca href=\"https:\u002F\u002Fx.ai\u002F\">xAI\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Fgrok.com\u002F\">Grok\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FWikipedia\">Wikipedia\u003C\u002Fa>, and the traffic\u002FSEO references that shaped the early reaction, like \u003Ca href=\"https:\u002F\u002Fwww.similarweb.com\u002F\">Similarweb\u003C\u002Fa>. The interesting part isn’t that Grokipedia exists. It’s that it exposes how fragile “AI-generated reference content” gets the moment you ship it into the open.\u003C\u002Fp>\u003Ch2>The launch story is already the warning label\u003C\u002Fh2>\u003Cblockquote>On October 6, 2025, Musk announced that the early version of Grokipedia was scheduled for release in two weeks, but the project was postponed briefly to address content quality issues.\u003C\u002Fblockquote>\u003Cp>What this actually means is simple: they were moving fast until the output looked bad enough that even they couldn’t pretend it was ready. I’ve seen this pattern in internal tools all the time. The demo works, the pipeline works, and then the first real content dump shows you how much cleanup the model still needs. Grokipedia launched on October 27, 2025 as “v 0.1,” which is basically a polite way of saying “please don’t judge this too hard yet.”\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779358589412-dwzr.png\" alt=\"Grokipedia turns AI wiki into a copy loop\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>The article says it launched with over 800,000 entries, which sounds huge until you compare it with English Wikipedia’s seven million-plus articles. That gap matters. A knowledge product is not impressive because it has a lot of pages. It’s impressive when the pages are reliable, maintainable, and traceable. If you’re building your own internal wiki or documentation system, I’d treat launch volume as a vanity metric. Quality delay plus versioned launch usually means the team knows the content layer is still unstable.\u003C\u002Fp>\u003Cp>I ran into this same issue when I tried to build a model-assisted docs generator for a product team. The first pass looked fine in a browser. Then we started checking edge cases, and the model quietly invented terminology, flattened nuance, and copied phrasing from source docs without preserving the actual constraints. The fix wasn’t “more model.” It was stricter source boundaries, explicit review stages, and a hard rule that generated pages could not publish without traceable provenance.\u003C\u002Fp>\u003Cp>How to apply it: if your system publishes factual content, make launch readiness depend on three things:\u003C\u002Fp>\u003Cul>\u003Cli>source provenance for every claim, not just a general citation block\u003C\u002Fli>\u003Cli>manual review for high-risk topics before anything goes public\u003C\u002Fli>\u003Cli>versioned releases so you can tell whether a bad page came from the model, the source corpus, or the review step\u003C\u002Fli>\u003C\u002Ful>\u003Cp>That’s the boring answer. It’s also the one that keeps you from shipping a confidence machine that lies with formatting.\u003C\u002Fp>\u003Ch2>Copying Wikipedia is not a shortcut, it’s the product\u003C\u002Fh2>\u003Cblockquote>Some articles are nearly identical to their Wikipedia entries, and some Grokipedia articles were republished almost verbatim, accompanied by a disclaimer noting that the content was “adapted from Wikipedia” under a Creative Commons license.\u003C\u002Fblockquote>\u003Cp>What this actually means is that Grokipedia is not starting from zero. It’s bootstrapping itself with the labor of an existing community and then trying to repackage the result as a different editorial system. That’s not inherently illegal when the license allows it, but it changes the product question from “can the model write?” to “what exactly is new here?”\u003C\u002Fp>\u003Cp>The article gives examples like AMD, Lamborghini, and PlayStation 5 being copied from Wikipedia. Musk later clarified that the duplication was intentional and that the team told Grok to compile Wikipedia’s top 1 million articles and make changes to them. That line matters more than the launch hype. It tells me the system was designed as a transform pipeline: ingest established reference content, rewrite it at scale, and then let the model decide what to keep.\u003C\u002Fp>\u003Cp>I’ve built content migration scripts that looked a lot like this. You take a trusted corpus, run transformations, and then compare the output. The danger is that the rewrite step becomes the editorial step. Once that happens, you’re no longer curating facts. You’re asking a model to preserve meaning while changing surface form, and that’s exactly where subtle errors creep in.\u003C\u002Fp>\u003Cp>If you’re applying this pattern in your own stack, I’d do it with a strict three-column view:\u003C\u002Fp>\u003Cul>\u003Cli>original source text\u003C\u002Fli>\u003Cli>machine-transformed draft\u003C\u002Fli>\u003Cli>human-reviewed final version\u003C\u002Fli>\u003C\u002Ful>\u003Cp>That makes it obvious where the drift happened. Without that separation, “adapted from” is just a nice label on top of a copy job.\u003C\u002Fp>\u003Cp>There’s also a licensing wrinkle here. Wikipedia-derived content carries \u003Ca href=\"https:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby-sa\u002F4.0\u002F\">CC BY-SA 4.0\u003C\u002Fa>, while Grokipedia’s non-Wikipedia content uses the \u003Ca href=\"https:\u002F\u002Fx.ai\u002Flegal\u002Fx-community-license\">X Community License\u003C\u002Fa>. If you’re building anything that mixes open content with proprietary model output, stop pretending the license is a footnote. It’s part of the architecture.\u003C\u002Fp>\u003Ch2>A model that reviews itself is still a model, not an editor\u003C\u002Fh2>\u003Cblockquote>Articles cannot be directly edited, though logged-in visitors to the encyclopedia can suggest new articles or corrections via a pop-up form, which are reviewed by Grok.\u003C\u002Fblockquote>\u003Cp>What this actually means is that Grokipedia replaced community editing with model-mediated intake. That sounds tidy until you remember that the model is both the writer and the reviewer. I don’t love that setup. It collapses two separate jobs into one probabilistic system and then asks users to trust the result because there’s a feedback form.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779358591701-bxg8.png\" alt=\"Grokipedia turns AI wiki into a copy loop\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>The article says that starting in version 0.2, Grok reviews and implements approved suggested edits, and a small panel rotates through recently edited articles. That’s a nice UI detail, but it doesn’t solve the core issue. A real editorial system needs disagreement, provenance, and traceable decisions. A model can simulate those things, but it can’t independently guarantee them.\u003C\u002Fp>\u003Cp>I’ve been burned by this in \u003Ca href=\"\u002Ftag\u002Fagent\">agent\u003C\u002Fa> workflows. You build a “review” step and think you’ve added control, but if the reviewer is the same model family, tuned the same way, on the same corpus, you’ve mostly added ceremony. The system still has one epistemic center. That’s the problem.\u003C\u002Fp>\u003Cp>How to apply it: if you want AI-assisted review, split the responsibilities hard:\u003C\u002Fp>\u003Cul>\u003Cli>one model or service generates drafts\u003C\u002Fli>\u003Cli>a different validator checks citations, dates, and named entities\u003C\u002Fli>\u003Cli>a human handles anything contentious, political, or legally sensitive\u003C\u002Fli>\u003C\u002Ful>\u003Cp>And if you can’t afford that split, don’t call it review. Call it auto-suggest. Being honest is cheaper than cleaning up a public mistake later.\u003C\u002Fp>\u003Ch2>The bias fight is really about source selection\u003C\u002Fh2>\u003Cblockquote>The xAI founder Elon Musk suggested Grokipedia could be an alternative to Wikipedia that would “purge out the propaganda” he believes is promoted by the latter, describing Wikipedia as “woke” and an “extension of legacy media propaganda”.\u003C\u002Fblockquote>\u003Cp>What this actually means is that Grokipedia was framed from the start as an ideological correction, not just a neutral encyclopedia. Once you say you’re removing “propaganda,” you’ve already chosen a side. The article also notes criticism that Grokipedia promotes right-wing perspectives and Musk’s views, and that coverage has tied it to debunked conspiracy theories on topics like HIV\u002FAIDS denialism, vaccines and autism, climate change, and race and intelligence.\u003C\u002Fp>\u003Cp>This is where people get slippery. They talk about “bias” like it’s a vague fog in the system. It isn’t. Bias shows up in source choice, phrasing, omission, and weighting. The article says studies found Grokipedia using low-credibility sources like X conversations and neo-Nazi websites, and writing about far-right figures in a promotional manner. That’s not a minor tuning issue. That’s a source policy failure.\u003C\u002Fp>\u003Cp>If you’re building a factual assistant, your source policy matters more than your prompt. I mean that literally. A model can only be as disciplined as the corpus and ranking rules around it. If you feed it junk, it will synthesize junk with better grammar. If you want less bias, you need source tiering, exclusion lists, and explicit topic-specific rules.\u003C\u002Fp>\u003Cp>My practical rule is this: for any public-facing knowledge product, I want three source buckets:\u003C\u002Fp>\u003Cul>\u003Cli>primary sources for claims that can be verified directly\u003C\u002Fli>\u003Cli>secondary sources for context and interpretation\u003C\u002Fli>\u003Cli>blocked sources for anything that is known to be unreliable or manipulative\u003C\u002Fli>\u003C\u002Ful>\u003Cp>And yes, you need to revisit those buckets often. Source quality decays. So does your confidence if you pretend it doesn’t.\u003C\u002Fp>\u003Ch2>Search traffic is not trust, it’s distribution\u003C\u002Fh2>\u003Cblockquote>According to an initial analysis of usage figures by Similarweb, Grokipedia recorded a peak of over 460,000 website visits in the US on October 28, 2025.\u003C\u002Fblockquote>\u003Cp>What this actually means is that launch attention is not the same thing as durable usage. The article says traffic later dropped to around 35,000 visits per day between November 8 and 11, 2025. That’s the kind of pattern you see when curiosity outruns habit. People check the thing out, then leave when the content quality or relevance doesn’t justify a return visit.\u003C\u002Fp>\u003Cp>That’s an important lesson for anyone shipping AI content at scale. You can get indexed, cited, and even surfaced in other AI systems without actually becoming trusted. The article notes that Grokipedia later appeared in GPT-5.2 responses and in \u003Ca href=\"\u002Ftag\u002Fgoogle\">Google\u003C\u002Fa>’s AI Overviews, AI Mode, \u003Ca href=\"\u002Ftag\u002Fgemini\">Gemini\u003C\u002Fa>, \u003Ca href=\"\u002Ftag\u002Fcopilot\">Copilot\u003C\u002Fa>, and Perplexity for niche factual questions. That’s distribution. It is not validation.\u003C\u002Fp>\u003Cp>I’ve watched teams confuse indexing with authority. They celebrate when the crawler shows up, then act surprised when the pages don’t convert into retention. Search engines and answer engines are not grading your epistemology. They’re just routing queries through whatever they can find.\u003C\u002Fp>\u003Cp>If you’re applying this to your own project, track two separate metrics:\u003C\u002Fp>\u003Cul>\u003Cli>discovery metrics like indexing, impressions, and referral traffic\u003C\u002Fli>\u003Cli>trust metrics like repeat usage, correction rate, and citation acceptance\u003C\u002Fli>\u003C\u002Ful>\u003Cp>If discovery rises while trust falls, you don’t have a growth story. You have a visibility problem wrapped around a quality problem.\u003C\u002Fp>\u003Ch2>The real product is a content pipeline, not an encyclopedia\u003C\u002Fh2>\u003Cblockquote>The Grok large language model generates and fact-checks articles on Grokipedia.\u003C\u002Fblockquote>\u003Cp>What this actually means is that Grokipedia is better understood as a content pipeline with an encyclopedia-shaped front end. That distinction matters. The user sees a search bar and article pages, but the actual product is the chain of generation, adaptation, review, licensing, and publication behind it.\u003C\u002Fp>\u003Cp>The article describes a minimalist design, a simple homepage, and a large search bar. That’s familiar territory. A lot of AI products hide complexity behind a clean surface because the surface is the easiest thing to ship. But if the pipeline underneath is noisy, the minimal UI just makes the mistakes feel more official.\u003C\u002Fp>\u003Cp>I like thinking about this in terms of failure modes. If the pipeline is the product, then every stage needs a testable contract. Generation needs citation rules. Review needs rejection criteria. Publication needs rollback. Search needs ranking controls. Without those, you’re not operating a reference system. You’re operating a text factory.\u003C\u002Fp>\u003Cp>Here’s the part I’d copy if I were building a smaller, saner version of this:\u003C\u002Fp>\u003Cul>\u003Cli>generate only from approved source sets\u003C\u002Fli>\u003Cli>store source spans alongside generated claims\u003C\u002Fli>\u003Cli>flag any uncited or weakly cited paragraph before publish\u003C\u002Fli>\u003Cli>keep a diff between source text and published text\u003C\u002Fli>\u003Cli>log every human override\u003C\u002Fli>\u003C\u002Ful>\u003Cp>That’s the difference between “AI encyclopedia” as branding and as engineering. One is a pitch. The other is a maintenance burden you can actually inspect.\u003C\u002Fp>\u003Ch2>The template you can copy\u003C\u002Fh2>\u003Cpre>\u003Ccode># AI encyclopedia content pipeline template\n\n## Goal\nBuild a reference site where AI can draft, rewrite, and review factual pages without hiding provenance.\n\n## Inputs\n- Approved primary sources\n- Approved secondary sources\n- Blocked sources list\n- Topic risk list\n- Human review queue\n\n## Workflow\n1. Ingest source material.\n2. Classify each source as primary, secondary, or blocked.\n3. Generate a draft only from approved sources.\n4. Attach source spans to every factual claim.\n5. Run a separate validation pass for:\n   - dates\n   - names\n   - numbers\n   - quoted text\n   - sensitive topics\n6. Route high-risk pages to human review.\n7. Publish only after validation passes.\n8. Store the source draft, transformed draft, reviewer notes, and final version.\n9. Log every post-publish correction.\n\n## Page schema\n- title\n- summary\n- claims[]\n- sources[]\n- source_spans[]\n- review_status\n- risk_level\n- last_updated\n- revision_history[]\n\n## Claim format\nEach claim should include:\n- claim_text\n- source_url\n- source_quote\n- confidence_level\n- reviewer_status\n\n## Review rules\n- Reject any claim without a source.\n- Reject any paragraph that depends on blocked sources.\n- Require human approval for political, medical, legal, or security topics.\n- Require a rollback path for every published revision.\n\n## Copy prompt for generation\nYou are drafting a factual encyclopedia entry.\nUse only the approved sources provided below.\nDo not invent facts.\nDo not infer dates, numbers, or names unless they are explicitly stated.\nFor every factual sentence, attach at least one source span.\nIf a claim is uncertain, mark it as uncertain instead of guessing.\n\n## Copy prompt for review\nYou are validating a factual encyclopedia draft.\nCheck every claim against the provided sources.\nReject unsupported claims.\nFlag weak citations.\nFlag ideological language.\nFlag promotional wording.\nReturn a list of accept, revise, or reject decisions with reasons.\n\n## Minimal publish checklist\n- [ ] Source coverage complete\n- [ ] No blocked sources used\n- [ ] High-risk topics reviewed by a human\n- [ ] Claims linked to source spans\n- [ ] Rollback tested\n- [ ] Revision log stored\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>This is not a clone of Grokipedia. It’s the part I’d actually want if I had to build a factual AI system that I didn’t want to babysit every day. The point is to separate generation from validation, keep provenance visible, and make the review step real instead of ceremonial.\u003C\u002Fp>\u003Cp>Source attribution: this breakdown is based on the \u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGrokipedia\">Wikipedia article on Grokipedia\u003C\u002Fa> and related sources it cites. My template is original, but the failure modes and workflow patterns are derived from the public reporting summarized there.\u003C\u002Fp>","I break down Grokipedia’s launch, bias, and copy-from-Wikipedia workflow, then give you a reusable AI-encyclopedia template.","en.wikipedia.org","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGrokipedia",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779358589412-dwzr.png","tools","en","5923250d-5aa4-48e5-ad91-9c9061e80d5a",[17,18,19,20,21],"Grokipedia","Wikipedia","xAI","AI knowledge base","source provenance",[23,24,25],"AI encyclopedias fail fast when generation and review live in the same model loop.","Copying a trusted corpus can bootstrap coverage, but it also imports licensing and provenance problems.","If you want factual AI content, source policy and rollback matter more than the model brand.",2,"2026-05-21T10:16:08.00123+00:00","2026-05-21T10:16:07.959+00:00","a7343b93-37cc-4634-a2bc-707f6275bdb6",{"tags":31,"relatedLang":42,"relatedPosts":46},[32,34,36,38,40],{"name":21,"slug":33},"source-provenance",{"name":19,"slug":35},"xai",{"name":20,"slug":37},"ai-knowledge-base",{"name":18,"slug":39},"wikipedia",{"name":17,"slug":41},"grokipedia",{"id":15,"slug":43,"title":44,"language":45},"grokipedia-turns-ai-wiki-into-copy-loop-zh","Grokipedia 把 AI 百科做成複製迴圈","zh",[47,53,59,65,71,77],{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":13},"1e0d71a2-19ae-44f4-970b-d27f77ad5a8a","nvidia-lg-ai-collaboration-playbook-en","Nvidia and LG turn AI plans into a playbook","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781056992194-i3tx.png","2026-06-10T02:02:46.922181+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":13},"9db77f6f-0d31-4686-86d9-16eb9615633d","ollama-best-free-ai-path-2026-en","Ollama is the best free AI path in 2026 for real work","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781056075632-qzpq.png","2026-06-10T01:47:25.10989+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":13},"c12c0470-eb29-4e44-872d-c133a84a1bc8","awesome-production-ml-turns-chaos-into-stack-en","This MLOps list turns chaos into a stack","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781055237524-86fa.png","2026-06-10T01:33:15.495884+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":13},"58924f21-83f4-405d-8d9a-4af334e9d030","bentoml-turns-model-serving-into-python-apis-en","BentoML turns model serving into Python APIs","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781054304942-bxxs.png","2026-06-10T01:17:56.721066+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":13},"aa96e422-2b01-4480-b4ce-a646be8e0993","magenta-realtime-2-score-inside-daw-en","Magenta RealTime 2 lets you score in the DAW","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781046208039-ksdz.png","2026-06-09T23:02:56.428086+00:00",{"id":78,"slug":79,"title":80,"cover_image":81,"image_url":81,"created_at":82,"category":13},"c79bca38-50b2-4d80-9a48-7f4d1afd051a","open-source-ai-tools-beat-claude-paid-tiers-en","Open-source AI tools beat Claude’s paid tiers on value","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781045269190-a1ow.png","2026-06-09T22:47:20.7972+00:00",[84,89,94,99,104,109,114,119,124,129],{"id":85,"slug":86,"title":87,"created_at":88},"8008f1a9-7a00-4bad-88c9-3eedc9c6b4b1","surepath-ai-mcp-policy-controls-en","SurePath AI's New MCP Policy Controls Enhance AI Security","2026-03-26T01:26:52.222015+00:00",{"id":90,"slug":91,"title":92,"created_at":93},"27e39a8f-b65d-4f7b-a875-859e2b210156","mcp-standard-ai-tools-2026-en","MCP Standard in 2026: Integrating AI Tools","2026-03-26T01:27:43.127519+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"165f9a19-c92d-46ba-b3f0-7125f662921d","rag-2026-transforming-enterprise-ai-en","How RAG in 2026 is Transforming Enterprise AI","2026-03-26T01:28:11.485236+00:00",{"id":100,"slug":101,"title":102,"created_at":103},"6a2a8e6e-b956-49d8-be12-cc47bdc132b2","mastering-ai-prompts-2026-guide-en","Mastering AI Prompts: A 2026 Guide for Developers","2026-03-26T01:29:07.835148+00:00",{"id":105,"slug":106,"title":107,"created_at":108},"3ab2c67e-4664-4c67-a013-687a2f605814","garry-tan-open-sources-claude-code-toolkit-en","Garry Tan Open-Sources a Claude Code Toolkit","2026-03-26T08:26:20.245934+00:00",{"id":110,"slug":111,"title":112,"created_at":113},"66a7cbf8-7e76-41d4-9bbf-eaca9761bf69","github-ai-projects-to-watch-in-2026-en","20 GitHub AI Projects to Watch in 2026","2026-03-26T08:28:09.752027+00:00",{"id":115,"slug":116,"title":117,"created_at":118},"9f332fda-eace-448a-a292-2283951eee71","practical-github-guide-learning-ml-2026-en","A Practical GitHub Guide to Learning ML in 2026","2026-03-27T01:16:50.125678+00:00",{"id":120,"slug":121,"title":122,"created_at":123},"1b1f637d-0f4d-42bd-974b-07b53829144d","aiml-2026-student-ai-ml-lab-repo-review-en","AIML-2026 Is a Bare-Bones Student Lab Repo","2026-03-27T01:21:51.661231+00:00",{"id":125,"slug":126,"title":127,"created_at":128},"6d1bf3f6-e191-4d30-b55b-8a0722fa6afe","ai-trending-github-repos-and-research-feeds-en","AI Trending Tracks Repos and Research Feeds","2026-03-27T01:31:35.709532+00:00",{"id":130,"slug":131,"title":132,"created_at":133},"010539a1-4c3a-4bd3-937a-26616422ee0d","awesome-ai-for-science-research-tools-map-en","Awesome AI for Science Is Becoming a Real Research Map","2026-03-27T01:46:50.89513+00:00"]