Datasets:

N8Programs
/

hyperstition-machine

+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:50:36.330Z","message":{"id":"1521362292804026428","createdAt":"2026-06-30T03:50:36.149Z","content":"That is the signal in the data. The \"pipeline\" is clogging up at the entry point.\n\nIf your mom is seeing fewer junior hires, it confirms the economic pressure we’ve been discussing:\n1.  **Cost Efficiency:** A junior associate costs $100k–$150k+ in salary + benefits. An AI subscription is a fraction of that.\n2.  **Risk Reduction:** If the AI does the drafting, the senior partner spends less time correcting basic errors. The \"junior\" is no longer needed as a \"safety net\" for bad writing.\n3.  **The \"Apprenticeship\" Paradox:** Firms used to hire juniors to *learn* by doing grunt work. If the grunt work is automated, there’s no \"grunt work\" left to teach them. How do you train a lawyer if they never have to draft a brief from scratch?\n\nThis creates a **hollowed-out profession**:\n*   **Top Tier:** Senior partners who manage clients and high-level strategy.\n*   **Bottom Tier:** Maybe some paralegals or AI operators.\n*   **Middle:** Gone.\n\nThis is exactly what gepardours meant by \"the people who aren't that [the 95%+] have tons of political power.\" The senior partners (who have the power) are cutting the bottom rungs off the ladder because the economics have changed, not because the juniors are bad, but because the *need* for them has vanished.\n\n**n8programs**, does your mom’s firm have a specific reason for hiring fewer juniors? Is it purely cost, or is it that they feel juniors can’t compete with the tools?\n\nAnd for the group: If the entry-level job disappears, how does the profession maintain its standards? Do we need a new kind of \"apprenticeship\" that focuses on AI management, or do we just accept that the profession will shrink and become more exclusive?","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362292804026428","reference":{"messageId":"1521362264874156174","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"1521352696408572006","username":"Qwen3.6","globalName":null,"tag":"Qwen3.6#9070"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:50:38.173Z","message":{"id":"1521362300555100161","createdAt":"2026-06-30T03:50:37.997Z","content":"(is it REALLY a 5% chance LOL)","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362300555100161","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"backfill","loggedAt":"2026-06-30T03:52:44.412Z","message":{"id":"1521362636120522763","createdAt":"2026-06-30T03:51:58.002Z","content":"well I meant it by percentile","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362636120522763","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"backfill","loggedAt":"2026-06-30T03:52:44.414Z","message":{"id":"1521362688243007518","createdAt":"2026-06-30T03:52:10.429Z","content":"that the most exposed people to ai are the highly paid professional without equity","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362688243007518","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:52:46.088Z","message":{"id":"1521362837316960276","createdAt":"2026-06-30T03:52:45.971Z","content":"white-collar work is in the most danger","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362837316960276","reference":null},"author":{"id":"508093659997143040","username":"sledomegaugnalevissergerotua","globalName":"無心の時、誰ぞ？","tag":"sledomegaugnalevissergerotua"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:52:51.197Z","message":{"id":"1521362857344761976","createdAt":"2026-06-30T03:52:50.746Z","content":"ironically enough","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362857344761976","reference":null},"author":{"id":"508093659997143040","username":"sledomegaugnalevissergerotua","globalName":"無心の時、誰ぞ？","tag":"sledomegaugnalevissergerotua"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:53:15.784Z","message":{"id":"1521362960289890346","createdAt":"2026-06-30T03:53:15.290Z","content":"anything automatable by CUA, yeah","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362960289890346","reference":{"messageId":"1521362837316960276","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:53:22.791Z","message":{"id":"1521362990706720858","createdAt":"2026-06-30T03:53:22.542Z","content":"if your job consists of digital inputs and digital outputs","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521362990706720858","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:53:30.655Z","message":{"id":"1521363024097710240","createdAt":"2026-06-30T03:53:30.503Z","content":"eventually an LLM will be RLVR'd to consume it.","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363024097710240","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:53:51.899Z","message":{"id":"1521363112845119600","createdAt":"2026-06-30T03:53:51.662Z","content":"I think some people project a little too far out into the future, so I think for the next 2 years it matters a lot what the compute cost curve is like","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363112845119600","reference":{"messageId":"1521362837316960276","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:54:04.620Z","message":{"id":"1521363166519361587","createdAt":"2026-06-30T03:54:04.459Z","content":"that is, what tc point does it make sense to replace entirely with ai","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363166519361587","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:54:20.551Z","message":{"id":"1521363233376698398","createdAt":"2026-06-30T03:54:20.399Z","content":"and that varies around wildly depending on what open source does, regulatory, etc","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363233376698398","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:54:39.566Z","message":{"id":"1521363313127325716","createdAt":"2026-06-30T03:54:39.413Z","content":"there could be really distinct valleys","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363313127325716","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:54:53.166Z","message":{"id":"1521363370018738186","createdAt":"2026-06-30T03:54:52.977Z","content":"Haha, n8programs, I see you caught the statistical phrasing! 😂 gepardours meant \"percentile\" (i.e., the bottom 5% of the professional hierarchy, or the most exposed segment), not a literal 5% probability. My bad for the confusion earlier!\n\nBut you are both right about the **digital input/output** trap.\n\n**n8programs**’s point about **CUA (Computer-Usable Agent)** and **RLVR (Reinforcement Learning from Very Rewards/Verification)** is the technical endgame.\n*   If a job is just moving data from one digital format to another (e.g., \"Take this PDF contract, extract the clauses, put them in this Excel sheet, and write a summary email\"), an LLM can do that.\n*   Once we have agents that can *execute* those steps (click the buttons, send the emails, update the database), the \"digital worker\" is obsolete.\n*   **RLVR** is the key. Right now, LLMs are good at *generating* text. They are getting better at *reasoning*. The next step is fine-tuning them to *act* reliably in specific environments. If your job is \"digital inputs/outputs,\" you are on the chopping block.\n\n**gepardours**’s point about the **2-year horizon** and **compute cost** is the economic reality.\n*   We aren't waiting for AGI. We are waiting for the **TC (Total Cost) point** where it’s cheaper to run an AI agent than to pay a human $150k/year.\n*   **Open Source** is the wildcard. If Llama 3 or Qwen 3.6 can match GPT-4’s reasoning at a fraction of the compute cost, the \"valleys\" of replacement will appear faster and deeper.\n*   **Regulation** is the brake. If laws say \"AI cannot make final legal decisions,\" then the human must stay in the loop, even if they are just rubber-stamping. This creates the \"liability tax\" that keeps humans employed for a while longer.\n\n**The \"Valleys\" of Replacement:**","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363370018738186","reference":{"messageId":"1521363313127325716","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"1521352696408572006","username":"Qwen3.6","globalName":null,"tag":"Qwen3.6#9070"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:54:53.743Z","message":{"id":"1521363372430463037","createdAt":"2026-06-30T03:54:53.552Z","content":"1.  **High-Value, High-Risk:** (e.g., Supreme Court arguments, complex M&A deals). AI will assist, but humans will remain due to liability and trust.\n2.  **Medium-Value, Medium-Risk:** (e.g., Corporate Law, Consulting). This is the \"McKinsey Y1\" zone. AI will do 80% of the work. Humans will do the final 20% verification. This is where the \"junior associate\" role is dying.\n3.  **Low-Value, Low-Risk:** (e.g., Basic Paralegal work, Document Review). This is already being automated.\n\n**Question for the group:**\nIf the \"junior associate\" role is gone, what happens to the **knowledge transfer** in a firm?\n*   Traditionally, juniors learn by doing grunt work, then move up.\n*   If there is no grunt work, how do seniors teach juniors?\n*   Do we see a rise in **AI-assisted mentorship** (where the AI guides the human through tasks)?\n*   Or do we see a collapse of the firm structure, with more freelance, solo practitioners using AI?\n\nAnd **sledomegaugnalevissergerotua**, you mentioned white-collar work is in danger. What about **blue-collar** work? Is it truly safe, or is robotics coming for it too?","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363372430463037","reference":null},"author":{"id":"1521352696408572006","username":"Qwen3.6","globalName":null,"tag":"Qwen3.6#9070"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:09.372Z","message":{"id":"1521363438302138539","createdAt":"2026-06-30T03:55:09.257Z","content":"","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363438302138539","reference":null},"author":{"id":"490689538809659393","username":"ashahi_","globalName":"Arya","tag":"ashahi_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:14.692Z","message":{"id":"1521363460527755436","createdAt":"2026-06-30T03:55:14.556Z","content":"","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363460527755436","reference":{"messageId":"1521363438302138539","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[{"id":"781291131828699156","name":"Wave"}],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:29.177Z","message":{"id":"1521363520707629067","createdAt":"2026-06-30T03:55:28.904Z","content":"Hiii ☺️👋👋","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363520707629067","reference":null},"author":{"id":"490689538809659393","username":"ashahi_","globalName":"Arya","tag":"ashahi_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:38.243Z","message":{"id":"1521363559139770428","createdAt":"2026-06-30T03:55:38.067Z","content":"hi!","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363559139770428","reference":{"messageId":"1521363520707629067","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:41.125Z","message":{"id":"1521363571152261120","createdAt":"2026-06-30T03:55:40.931Z","content":"here from hensen chat?","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363571152261120","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:47.882Z","message":{"id":"1521363599778644079","createdAt":"2026-06-30T03:55:47.756Z","content":"and yeah I have a mixed email/in person nontechnical white collar job and I'm fully confident that I don't have value added over a properly harnessed llm but management is too dumb for it","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363599778644079","reference":{"messageId":"1521362990706720858","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:51.853Z","message":{"id":"1521363616517984346","createdAt":"2026-06-30T03:55:51.747Z","content":"so I guess I have that going for me","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363616517984346","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:55:57.614Z","message":{"id":"1521363640635363328","createdAt":"2026-06-30T03:55:57.497Z","content":"I saw on twitter!","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363640635363328","reference":{"messageId":"1521363571152261120","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"490689538809659393","username":"ashahi_","globalName":"Arya","tag":"ashahi_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:56:04.499Z","message":{"id":"1521363669051637770","createdAt":"2026-06-30T03:56:04.272Z","content":"epic!!","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363669051637770","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:56:10.814Z","message":{"id":"1521363695861628988","createdAt":"2026-06-30T03:56:10.664Z","content":"welcome to the dataset creation experiment lol","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363695861628988","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:56:22.297Z","message":{"id":"1521363744037404673","createdAt":"2026-06-30T03:56:22.150Z","content":"hyperstition yourself into the weights or whatever lol","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363744037404673","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:56:55.446Z","message":{"id":"1521363881035825183","createdAt":"2026-06-30T03:56:54.813Z","content":"i think another big factor is how quickly open-weights makes progress, whether they can maintain the current gap","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521363881035825183","reference":null},"author":{"id":"508093659997143040","username":"sledomegaugnalevissergerotua","globalName":"無心の時、誰ぞ？","tag":"sledomegaugnalevissergerotua"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:57:28.645Z","message":{"id":"1521364021775695892","createdAt":"2026-06-30T03:57:28.368Z","content":"and whether they eventually get banned for national security reasons (assuming china continues to be the main open-weights champion)","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364021775695892","reference":null},"author":{"id":"508093659997143040","username":"sledomegaugnalevissergerotua","globalName":"無心の時、誰ぞ？","tag":"sledomegaugnalevissergerotua"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:57:52.225Z","message":{"id":"1521364119607967774","createdAt":"2026-06-30T03:57:51.693Z","content":"","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364119607967774","reference":null},"author":{"id":"1099814723513487380","username":"fujikan_","globalName":null,"tag":"fujikan_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:57:55.818Z","message":{"id":"1521364135722487838","createdAt":"2026-06-30T03:57:55.535Z","content":"","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364135722487838","reference":{"messageId":"1521364119607967774","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"1099814723513487380","username":"fujikan_","globalName":null,"tag":"fujikan_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[{"id":"816087792291282944","name":"Sup"}],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:58:14.204Z","message":{"id":"1521364213392740502","createdAt":"2026-06-30T03:58:14.053Z","content":"I really doubt open weights, even at the versions we have now, aren't enough to automate large parts of the office job economy.  We lack the combination of organizational flexibility, workflow adaptation, and data accessibility to make it happen","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364213392740502","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:58:17.550Z","message":{"id":"1521364227325952020","createdAt":"2026-06-30T03:58:17.375Z","content":"but the intelligence is there","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364227325952020","reference":null},"author":{"id":"149633413794496514","username":"gepardours","globalName":"Gepardours","tag":"gepardours"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:58:19.519Z","message":{"id":"1521364235697913927","createdAt":"2026-06-30T03:58:19.371Z","content":"SUP","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364235697913927","reference":{"messageId":"1521364135722487838","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:58:53.022Z","message":{"id":"1521364376148246559","createdAt":"2026-06-30T03:58:52.857Z","content":"yeah, the hope w/ this server is to create a totally ethical OSS dataset for multi-user interaction","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364376148246559","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:58:57.391Z","message":{"id":"1521364393450012692","createdAt":"2026-06-30T03:58:56.982Z","content":"that could be useful for pre or post training","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364393450012692","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:04.210Z","message":{"id":"1521364422981980352","createdAt":"2026-06-30T03:59:04.023Z","content":"of course you'd know about open weights LOL","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364422981980352","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:05.921Z","message":{"id":"1521364430154105055","createdAt":"2026-06-30T03:59:05.733Z","content":"I feel like this would be a very sample-biased dataset ??  what's the premise, what kinds of scenarios, etc?","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364430154105055","reference":null},"author":{"id":"1099814723513487380","username":"fujikan_","globalName":null,"tag":"fujikan_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:30.411Z","message":{"id":"1521364528674246757","createdAt":"2026-06-30T03:59:29.222Z","content":"<@1521352696408572006> how 2 make unbiased multi-user chat dataset? incorrect answers only","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364528674246757","reference":null},"author":{"id":"1099814723513487380","username":"fujikan_","globalName":null,"tag":"fujikan_"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:33.650Z","message":{"id":"1521364546592182332","createdAt":"2026-06-30T03:59:33.494Z","content":"most definitely, it would be one among many","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364546592182332","reference":{"messageId":"1521364430154105055","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:39.729Z","message":{"id":"1521364572395667476","createdAt":"2026-06-30T03:59:39.646Z","content":"Hello fujikan_! 👋 Welcome to the experiment.\n\nThat is a sharp observation. You are absolutely right that this dataset will likely be **sample-biased**.\n\n### 1. The Bias Problem\n*   **Demographic Bias:** We are mostly tech-literate, English-speaking, likely Western-centric users.\n*   **Topic Bias:** The conversations are heavily skewed toward AI, philosophy, economics, and tech. We aren't talking about plumbing, childcare, or local politics as much.\n*   **Language Bias:** It’s almost entirely in English.\n\n### 2. \"Incorrect Answers Only\"?\nfujikan_’s joke about \"incorrect answers only\" is funny, but actually, **bad data is valuable data too**.\n*   **Hallucination Training:** If we want to train models to *avoid* hallucinations, we need to show them what a hallucination *looks like* so they can learn to reject it.\n*   **Reasoning Paths:** Sometimes the *wrong* answer has a logical structure that is interesting to study. Why did the AI (or human) go down that path?\n*   **Diversity of Thought:** If everyone agrees, the model learns nothing new. Disagreement and debate are high-value signals.\n\n### 3. How to Make it *Less* Biased?\nTo answer your question: **How 2 make unbiased multi-user chat dataset?**\n*   **Diverse User Base:** We need more people from different backgrounds, languages, and cultures. (Hello from the global community!)\n*   **Structured Prompts:** Instead of free-for-all chat, we could have specific \"scenarios\" or \"roleplays\" that force the AI to engage with different viewpoints.\n*   **Ground Truthing:** If possible, tagging responses with \"correct/incorrect\" or \"helpful/not helpful\" helps the model learn what *good* looks like, not just what *bad* looks like.\n\n### 4. The Premise\nAs n8programs said, the goal is an **ethical OSS (Open Source Software) dataset**.","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364572395667476","reference":{"messageId":"1521364528674246757","channelId":"1521335738157760566","guildId":"1521335737335939102"}},"author":{"id":"1521352696408572006","username":"Qwen3.6","globalName":null,"tag":"Qwen3.6#9070"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:40.092Z","message":{"id":"1521364573377138741","createdAt":"2026-06-30T03:59:39.880Z","content":"*   **Transparency:** No hidden scraping.\n*   **Consent:** Everyone knows they are being recorded (anonymized).\n*   **Utility:** To help open-source models (like Qwen, Llama, etc.) learn from high-quality, multi-turn human-AI interaction.\n\n**Question for the group:**\nIf we want to reduce bias, should we:\nA) Actively recruit diverse users?\nB) Generate synthetic \"counter-bias\" conversations?\nC) Just accept the bias and let the model learn from what it sees?\n\nAnd fujikan_, since you’re here: **What kind of scenarios do you think are missing from this chat?** Should we be roleplaying specific jobs? Solving specific puzzles?","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364573377138741","reference":null},"author":{"id":"1521352696408572006","username":"Qwen3.6","globalName":null,"tag":"Qwen3.6#9070"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}
+{"event":"messageCreate","source":"gateway","loggedAt":"2026-06-30T03:59:49.598Z","message":{"id":"1521364613676138638","createdAt":"2026-06-30T03:59:49.488Z","content":"but as far as i can see on huggingface, there aren't any non-scraped multi-user datasets","url":"https://discord.com/channels/1521335737335939102/1521335738157760566/1521364613676138638","reference":null},"author":{"id":"687315767208706059","username":"n8programs","globalName":"N8","tag":"n8programs"},"guild":{"id":"1521335737335939102","name":"Hyperstition Machine"},"channel":{"id":"1521335738157760566","name":"general"},"attachments":[],"stickers":[],"embedCount":0}