{"id":3272,"slug":"llm","name":"LLM","type":"concept","created_at":"2026-08-08T04:52:50.069264+00:00","hidden":0,"aliases":["LLMs"],"episodes":[{"id":"396882e7-1305-440d-8299-505fc3b49f11","slug":"s21e14-just-the-links","subject":"s21e14: Just the Links","title":"Just the Links","publish_date":"2026-07-29T02:25:57+00:00","year":2026,"month":7,"season":21,"episode":14,"episode_label":"s21e14","canonical_url":"https://newsletter.danhon.com/archive/s21e14-just-the-links/","is_premium":0,"word_count":1168,"context":" … I wrote that! Not an <mark>LLM</mark>! Ugh, I feel gross.\n\nFIRST.\n\nArs Technica: A missing underscore sent innocent man to prison for 18 months\n1\nis a story … ","reading_minutes":6},{"id":"27d1b094-3cbd-4626-9fb2-83aaf118d33d","slug":"s21e13-the-irritating-aspect-of-spiciness","subject":"s21e13: The Irritating Aspect of Spiciness; Improving Pharmacies Somewhat","title":"The Irritating Aspect of Spiciness; Improving Pharmacies Somewhat","publish_date":"2026-07-27T23:58:52+00:00","year":2026,"month":7,"season":21,"episode":13,"episode_label":"s21e13","canonical_url":"https://newsletter.danhon.com/archive/s21e13-the-irritating-aspect-of-spiciness/","is_premium":0,"word_count":3035,"context":" … It is weird to people because computers are supposed to be programmable and “smart” but the thing to understand here is that when you ask an <mark>LLM</mark> to … ","reading_minutes":15},{"id":"475f0467-dd6a-4da1-9043-3a5bf4743cc2","slug":"s21e12-for-all-humanity-gordon-moore-folds-your","subject":"s21e12: For All Humanity?; Gordon Moore Folds Your Clothes","title":"For All Humanity?; Gordon Moore Folds Your Clothes","publish_date":"2026-07-22T03:07:03+00:00","year":2026,"month":7,"season":21,"episode":12,"episode_label":"s21e12","canonical_url":"https://newsletter.danhon.com/archive/s21e12-for-all-humanity-gordon-moore-folds-your/","is_premium":0,"word_count":1327,"context":" … Either <mark>LLM</mark> model weights aren't copyrightable, in which case distilling an <mark>LLM</mark> cannot be copyright infringement; or they are the product of mass copyright infringement themselves, in … ","reading_minutes":7},{"id":"5a97efa3-08c5-453c-9212-1ba7c28fcd99","slug":"s21e09-a-different-strategy-the-sanest-choice-in","subject":"s21e09: A Different Strategy, The Sanest Choice In An Insane World","title":"A Different Strategy, The Sanest Choice In An Insane World","publish_date":"2026-07-03T22:43:48+00:00","year":2026,"month":7,"season":21,"episode":9,"episode_label":"s21e09","canonical_url":"https://newsletter.danhon.com/archive/s21e09-a-different-strategy-the-sanest-choice-in/","is_premium":0,"word_count":1474,"context":" … In an insane world, it was the sanest choice.\n\n(Unlike contemporary <mark>LLMs</mark>, there’s not that much evidence that Arnold the Terminator would encourage John to die by … ","reading_minutes":7},{"id":"edeefbbc-74e0-4160-9161-e382c64ccbef","slug":"s21e07-anything-llms-can-do-i-can-do-better","subject":"s21e07: Anything LLMs Can Do, I Can Do Better","title":"Anything LLMs Can Do, I Can Do Better","publish_date":"2026-06-24T22:15:56+00:00","year":2026,"month":6,"season":21,"episode":7,"episode_label":"s21e07","canonical_url":"https://newsletter.danhon.com/archive/s21e07-anything-llms-can-do-i-can-do-better/","is_premium":0,"word_count":2072,"context":" … I won’t say <mark>LLMs</mark> can’t do that because one should never say never, and I also won’t say <mark>LLMs</mark> can’t do that because anything … ","reading_minutes":10},{"id":"85e15f08-3d3c-4013-8d12-39efd4f98c7a","slug":"s21e05-that-whole-moltbotopenclaw-yoloing-ai","subject":"s21e05: That Whole Moltbot/OpenClaw YOLOing AI Agents Thing","title":"That Whole Moltbot/OpenClaw YOLOing AI Agents Thing","publish_date":"2026-02-05T19:34:52+00:00","year":2026,"month":2,"season":21,"episode":5,"episode_label":"s21e05","canonical_url":"https://newsletter.danhon.com/archive/s21e05-that-whole-moltbotopenclaw-yoloing-ai/","is_premium":0,"word_count":4202,"context":" … But it’s because the <mark>LLM</mark> isn’t actually running a program (which can be good at counting!), it is, well, being an <mark>LLM</mark> about it.\n\nThis is … ","reading_minutes":21},{"id":"f6db1c46-155d-400c-87f6-7d83e3c76df8","slug":"s21e03-the-problem-is-defining-the-problem-the","subject":"s21e03: The Problem is Defining the Problem; The Unreasonable Effectiveness of Caring","title":"The Problem is Defining the Problem; The Unreasonable Effectiveness of Caring","publish_date":"2026-01-15T16:00:00+00:00","year":2026,"month":1,"season":21,"episode":3,"episode_label":"s21e03","canonical_url":"https://newsletter.danhon.com/archive/s21e03-the-problem-is-defining-the-problem-the/","is_premium":0,"word_count":3575,"context":" … And that's quite the opposite of what <mark>LLM</mark> assistants promise.\n3\n\nI think the point here is that agentic coding/<mark>LLM</mark>-assisted coding (and therefore I think … ","reading_minutes":18},{"id":"2ee82c63-615e-4cac-b767-26b0396d04b4","slug":"s21e01-things-could-have-been-different-llms-as","subject":"s21e01: Things Could Have Been Different; LLMs as LitTech, Their Use as Response to Mass Functional Illiteracy","title":"Things Could Have Been Different; LLMs as LitTech, Their Use as Response to Mass Functional Illiteracy","publish_date":"2026-01-07T18:32:44+00:00","year":2026,"month":1,"season":21,"episode":1,"episode_label":"s21e01","canonical_url":"https://newsletter.danhon.com/archive/s21e01-things-could-have-been-different-llms-as/","is_premium":0,"word_count":2917,"context":" … 1.1 <mark>LLMs</mark> as LitTech, Their Use as Response to Mass Functional Illiteracy\n\nI will start with some of the interesting bits first to get your attention before … ","reading_minutes":15},{"id":"81b66693-a3e6-4fd6-afb1-0932c73ab6e1","slug":"s20e09-an-end-of-year-opinion-about-ai-because","subject":"s20e09: An End Of Year Opinion About AI Because Why Not; Good Enough Mitigation of Reasonably Foreseeable Harm","title":"An End Of Year Opinion About AI Because Why Not; Good Enough Mitigation of Reasonably Foreseeable Harm","publish_date":"2025-12-29T21:58:35+00:00","year":2025,"month":12,"season":20,"episode":9,"episode_label":"s20e09","canonical_url":"https://newsletter.danhon.com/archive/s20e09-an-end-of-year-opinion-about-ai-because/","is_premium":0,"word_count":2783,"context":" … So I’ll just be quick and pretend that we’ve covered that “AI” can mean lots of things, “generative AI” is something specific, people are using <mark>LLMs</mark> … ","reading_minutes":14},{"id":"755f7fd9-60df-4c92-bd96-9867cd750a87","slug":"s20e08-efficiency-gains-i-understand-you-but","subject":"s20e08: Efficiency Gains; I Understand You, But Could Never Be You; Can You Teach A Machine To Be Human?","title":"Efficiency Gains; I Understand You, But Could Never Be You; Can You Teach A Machine To Be Human?","publish_date":"2025-09-04T00:16:03+00:00","year":2025,"month":9,"season":20,"episode":8,"episode_label":"s20e08","canonical_url":"https://newsletter.danhon.com/archive/s20e08-efficiency-gains-i-understand-you-but/","is_premium":0,"word_count":2523,"context":" … 1.0 Some Things That Caught My Attention\n\n1.1 Efficiency Gains\n\nIt is a bit of a cheap shot to say that <mark>LLMs</mark> aren’t great. That … ","reading_minutes":13},{"id":"060847e5-bf40-4387-96da-c14590528513","slug":"s20e06-not-about-how-it-looks","subject":"s20e06: Not About How it Looks","title":"Not About How it Looks","publish_date":"2025-08-26T21:02:44+00:00","year":2025,"month":8,"season":20,"episode":6,"episode_label":"s20e06","canonical_url":"https://newsletter.danhon.com/archive/s20e06-not-about-how-it-looks/","is_premium":0,"word_count":2195,"context":" … Maybe Grok (or whichever <mark>LLM</mark>) is less biased, or knows about more experts in whichever field? My gut feel is that Grok might “know” more experts, but that … ","reading_minutes":11},{"id":"f3edae0e-7ae0-468e-89d9-ef4228709a44","slug":"se20e05-ai-models-the-all-souls-exam-melvynn","subject":"se20e05: AI Models; The All Souls Exam; Melvynn Bragg as an Oxford Don; Grounding; Being Human; Worry About The Same Things","title":"se20e05: AI Models; The All Souls Exam; Melvynn Bragg as an Oxford Don; Grounding; Being Human; Worry About The Same Things","publish_date":"2025-08-19T17:19:28+00:00","year":2025,"month":8,"season":20,"episode":5,"episode_label":"s20e05","canonical_url":"https://newsletter.danhon.com/archive/se20e05-ai-models-the-all-souls-exam-melvynn/","is_premium":0,"word_count":2970,"context":" … 1.0 Some Things That Caught My Attention\n\n1.1 Should Anonymous Posting Online Be Forbidden?\n\nI wrote back in s20e01\n1\nabout how current <mark>LLMs</mark>/token predictors … ","reading_minutes":15},{"id":"ef4c89c6-be05-4cc3-8be0-6f8b2ca0fb85","slug":"s20e01-better-than-average-how-people-work","subject":"s20e01: Better than Average; How People Work","title":"Better than Average; How People Work","publish_date":"2025-08-08T22:33:00+00:00","year":2025,"month":8,"season":20,"episode":1,"episode_label":"s20e01","canonical_url":"https://newsletter.danhon.com/archive/s20e01-better-than-average-how-people-work/","is_premium":0,"word_count":2605,"context":" … posted an observation a few weeks back that went a bit like this:\n\n<mark>LLMs</mark> / generative AI / token prediction machines are trained on a mass of data\n\nThey are … ","reading_minutes":13},{"id":"a5dc4d35-2cc3-4a1f-966d-eab4644094c8","slug":"s19e11-your-model-is-naive-and-its-not-the","subject":"s19e11: Your Model is Naive and it’s Not The Territory","title":"Your Model is Naive and it’s Not The Territory","publish_date":"2025-05-30T18:21:31+00:00","year":2025,"month":5,"season":19,"episode":11,"episode_label":"s19e11","canonical_url":"https://newsletter.danhon.com/archive/s19e11-your-model-is-naive-and-its-not-the/","is_premium":0,"word_count":1523,"context":" … The models are complex enough when encoded in legislation and then a business rules engine, but now the models are potentially implicitly encoded into your <mark>LLM</mark> by you … ","reading_minutes":8},{"id":"b2f95417-dc59-49e6-a44f-50c0a51dbffa","slug":"s19e01-do-reply-use-plain-language-and-tell-the","subject":"s19e01: Do Reply; Use plain language, and tell the truth","title":"Do Reply; Use plain language, and tell the truth","publish_date":"2024-08-29T20:04:54.015071+00:00","year":2024,"month":8,"season":19,"episode":1,"episode_label":"s19e01","canonical_url":"https://newsletter.danhon.com/archive/s19e01-do-reply-use-plain-language-and-tell-the/","is_premium":0,"word_count":1374,"context":" … The reason why <mark>LLM</mark> text for me is bad is that it’s insipid, which is not a plain language word to use, but the secret is to … ","reading_minutes":7},{"id":"e7d753ff-34be-443c-8377-181fd85994e7","slug":"s18e07-do-not-reply-stickers-computers-were-a-bad","subject":"s18e07: Do Not Reply Stickers; Computers were a Bad Idea; Two Things About Startups; and more tidbits","title":"Do Not Reply Stickers; Computers were a Bad Idea; Two Things About Startups; and more tidbits","publish_date":"2024-04-29T18:18:07.335950+00:00","year":2024,"month":4,"season":18,"episode":7,"episode_label":"s18e07","canonical_url":"https://newsletter.danhon.com/archive/s18e07-do-not-reply-stickers-computers-were-a-bad/","is_premium":0,"word_count":1260,"context":" … voice-based <mark>LLM</mark> interfaces.\n\n(Of which! Voice-based <mark>LLM</mark> interfaces! Such a hot mess right now! A lot of the problems in voice interfaces like lack of discoverability … ","reading_minutes":6},{"id":"5e3ffcbe-d4fb-4f2a-88e5-1f38fe3e1f71","slug":"s18e04-what-are-we-doing-here-exactly-llms-are","subject":"s18e04: What Are We Doing Here, Exactly?; LLMs Are Playful, Actually","title":"What Are We Doing Here, Exactly?; LLMs Are Playful, Actually","publish_date":"2024-03-27T21:34:02.502635+00:00","year":2024,"month":3,"season":18,"episode":4,"episode_label":"s18e04","canonical_url":"https://newsletter.danhon.com/archive/s18e04-what-are-we-doing-here-exactly-llms-are/","is_premium":0,"word_count":3553,"context":" … 1.2 <mark>LLMs</mark> are Playful, Actually\n\nOh jeez where to begin with this one-slash-what a doozy.\n\nTwo things coming together in my mind here. Here’s … ","reading_minutes":18},{"id":"8dfb03ad-5e93-41c2-9dfb-2db2d06c694f","slug":"s17e12-justice-in-forensic-algorithms-the-one-5114","subject":"s17e12: Justice in Forensic Algorithms; The One With Retroactive Bereavement Fare Claims; Bits and Pieces","title":"Justice in Forensic Algorithms; The One With Retroactive Bereavement Fare Claims; Bits and Pieces","publish_date":"2024-02-16T17:48:20.171130+00:00","year":2024,"month":2,"season":17,"episode":12,"episode_label":"s17e12","canonical_url":"https://newsletter.danhon.com/archive/s17e12-justice-in-forensic-algorithms-the-one-5114/","is_premium":0,"word_count":1955,"context":" … Here’s the good bits, which is mainly me doing a précis (sorry, organic, sentient <mark>LLM</mark> summarization) of the paras 24-32 of the judgment:\n\nThe judge helpfully … ","reading_minutes":10},{"id":"0f403938-db93-4e49-a3a0-4cb3e5d8ab3e","slug":"s17e11-context-is-king-explicit-people","subject":"s17e11: Context is King; Explicit People Personality Preferences; OS Licensing, but Different","title":"Context is King; Explicit People Personality Preferences; OS Licensing, but Different","publish_date":"2024-02-14T23:32:27.147055+00:00","year":2024,"month":2,"season":17,"episode":11,"episode_label":"s17e11","canonical_url":"https://newsletter.danhon.com/archive/s17e11-context-is-king-explicit-people/","is_premium":0,"word_count":2828,"context":" … copy to be!\n\nI guess people will create the different versions with <mark>LLMs</mark>. (BAD <mark>LLM</mark>! NOW DO IT 10% LESS PUMPINGLY) Now I’m sad.\n\n1.2 OS … ","reading_minutes":14},{"id":"f1488de1-db96-4729-a480-4fa3ca7fdac4","slug":"s17e09-what-does-it-take-to-trust-the-young-ladys","subject":"s17e09: What does it take to trust?; The Young Lady’s Illustrated Tour Guide","title":"What does it take to trust?; The Young Lady’s Illustrated Tour Guide","publish_date":"2024-02-13T03:38:55.061867+00:00","year":2024,"month":2,"season":17,"episode":9,"episode_label":"s17e09","canonical_url":"https://newsletter.danhon.com/archive/s17e09-what-does-it-take-to-trust-the-young-ladys/","is_premium":0,"word_count":2423,"context":" … That time it was about how the Rabbit R1 hardware-enclosed <mark>LLM</mark> assistant had struck peoples’ attention, and whether the demoware showed actually-useful, real-world examples.\n\nAnyway … ","reading_minutes":12},{"id":"b0218499-69a1-41ef-8264-4f32fbf05bb8","slug":"s17e06-its-not-a-hardware-or-software-problem","subject":"s17e06: It’s not a hardware or software problem, dummy","title":"It’s not a hardware or software problem, dummy","publish_date":"2024-02-01T17:06:32.410126+00:00","year":2024,"month":2,"season":17,"episode":6,"episode_label":"s17e06","canonical_url":"https://newsletter.danhon.com/archive/s17e06-its-not-a-hardware-or-software-problem/","is_premium":0,"word_count":1059,"context":" … Rabbit and <mark>LLMs</mark> are a kind of symptom of this because they show the promise -- whether realized or not -- of being able to take tedious, siloed tasks that … ","reading_minutes":5},{"id":"fba24546-eb86-4b58-accc-f96b9938a72c","slug":"s17e01-how-old-things-keep-living-impossible-ai","subject":"s17e01: How old things keep living; Impossible AI; Look, just call it AI","title":"How old things keep living; Impossible AI; Look, just call it AI","publish_date":"2024-01-09T00:15:16.030093+00:00","year":2024,"month":1,"season":17,"episode":1,"episode_label":"s17e01","canonical_url":"https://newsletter.danhon.com/archive/s17e01-how-old-things-keep-living-impossible-ai/","is_premium":0,"word_count":3107,"context":" … Case Studies | Artificial Intelligence | JAMA Pediatrics | JAMA Network\n(\narchive.is\n)\n↩︎\n\nWhat I Learned Using Private <mark>LLMs</mark> to Write an Undergraduate History Essay – zwischenzugs\n(\narchive.is\n), 27 December, 2023\n↩︎","reading_minutes":16},{"id":"b34bf760-9852-40fc-a338-3c33dd1b3732","slug":"s16e16-not-up-and-to-the-right-just-deeper-really","subject":"s16e16: Not Up and To The Right; Just Deeper, Really; Better Wording Faster; Tools for Remembering; A Centaur for Caring?; Of Course They're Art","title":"Not Up and To The Right; Just Deeper, Really; Better Wording Faster; Tools for Remembering; A Centaur for Caring?; Of Course They're Art","publish_date":"2023-10-19T19:35:01.799791+00:00","year":2023,"month":10,"season":16,"episode":16,"episode_label":"s16e16","canonical_url":"https://newsletter.danhon.com/archive/s16e16-not-up-and-to-the-right-just-deeper-really/","is_premium":0,"word_count":3049,"context":" … Flynn, continuing my conversation about how <mark>LLMs</mark> and other so-called artificial intelligences can be used by people.\n\nThere was this concept of caring and relationship management. When … ","reading_minutes":15},{"id":"dd214afe-1999-4943-8e9d-f1165b2d92cd","slug":"s16e14-the-torment-nexus-section-rfps-man-rfps","subject":"s16e14: The Torment Nexus Section; RFPs, man. RFPs.","title":"The Torment Nexus Section; RFPs, man. RFPs.","publish_date":"2023-10-13T22:54:40.784297+00:00","year":2023,"month":10,"season":16,"episode":14,"episode_label":"s16e14","canonical_url":"https://newsletter.danhon.com/archive/s16e14-the-torment-nexus-section-rfps-man-rfps/","is_premium":0,"word_count":1815,"context":" … OK, so Wired did a\nreally\ninteresting / horrifying story about how <mark>LLMs</mark> (automatic text generation engines and word/phrase-relationship engines) have upended the whole business of responding … ","reading_minutes":9},{"id":"af3f5020-f20d-4c2e-a2f7-542b1a5528fa","slug":"s16e06-tech-journalism-but-not-tech-journalism","subject":"s16e06: Tech Journalism, But Not Tech Journalism; Sounds More Like A You Problem","title":"Tech Journalism, But Not Tech Journalism; Sounds More Like A You Problem","publish_date":"2023-10-02T20:48:49.598074+00:00","year":2023,"month":10,"season":16,"episode":6,"episode_label":"s16e06","canonical_url":"https://newsletter.danhon.com/archive/s16e06-tech-journalism-but-not-tech-journalism/","is_premium":0,"word_count":2469,"context":" … Just the other week, The Atlantic published a tool for searching the Books3 dataset\n[4]\nused to train generative <mark>LLMs</mark>, although the not-so-great news is that … ","reading_minutes":12},{"id":"69ea2a30-7bf0-4fa9-93d9-6614109ae087","slug":"s15e01-marks-on-a-flat-plane","subject":"s15e01: Marks on Flat Planes","title":"Marks on Flat Planes","publish_date":"2023-06-08T18:01:41.339766+00:00","year":2023,"month":6,"season":15,"episode":1,"episode_label":"s15e01","canonical_url":"https://newsletter.danhon.com/archive/s15e01-marks-on-a-flat-plane/","is_premium":0,"word_count":3502,"context":" … Fine-tune your <mark>LLM</mark> character-based model on all your Sherlock Holmes data so that when you see your dailies of your Sherlock Holmes piece of\ncontent\n, you … ","reading_minutes":18},{"id":"ee10c2d4-cab6-4fdd-8a17-7de86a6e2cca","slug":"s14e14-someone-comes-to-town-someone-leaves-town","subject":"s14e14: Someone Comes to Town, Someone Leaves Town; Streaks Broken; Aigg on your Face, Big Disgrace","title":"Someone Comes to Town, Someone Leaves Town; Streaks Broken; Aigg on your Face, Big Disgrace","publish_date":"2023-02-08T21:22:48.637242+00:00","year":2023,"month":2,"season":14,"episode":14,"episode_label":"s14e14","canonical_url":"https://newsletter.danhon.com/archive/s14e14-someone-comes-to-town-someone-leaves-town/","is_premium":0,"word_count":2153,"context":" … One reason why, I'm pretty sure, is that <mark>LLM</mark>-generated text is super easy to glaze over entirely because it's so predictable\nand we're already … ","reading_minutes":11},{"id":"57fd54db-3330-45db-9df6-786f0a8d9f56","slug":"s14e11-to-help-you-best-id-need-to-know-you","subject":"s14e11: To help you best, I'd need to know you","title":"To help you best, I'd need to know you","publish_date":"2023-01-27T20:05:00.756213+00:00","year":2023,"month":1,"season":14,"episode":11,"episode_label":"s14e11","canonical_url":"https://newsletter.danhon.com/archive/s14e11-to-help-you-best-id-need-to-know-you/","is_premium":0,"word_count":882,"context":" … Henley says the potential of <mark>LLMs</mark> is that they\n\ncould feed the relevant context to the model behind the scenes and use that to preemptively suggest what I … ","reading_minutes":4},{"id":"407942a5-5017-4cfe-82e9-6b8fe5fd0221","slug":"s12e37-this-was-not-the-future-we-were-promisedw","subject":"s12e37: This was not the future we were promised^w sold","title":"This was not the future we were promised^w sold","publish_date":"2022-07-12T21:27:07.165814+00:00","year":2022,"month":7,"season":12,"episode":37,"episode_label":"s12e37","canonical_url":"https://newsletter.danhon.com/archive/s12e37-this-was-not-the-future-we-were-promisedw/","is_premium":0,"word_count":1181,"context":" … Though engineer seems too formal, it immediately conveys the point I’m making: “Prompt engineers use natural language prompts and <mark>LLMs</mark> to produce new content”\n@AndrewKemendo\n\nCaught my … ","reading_minutes":6}]}