2

Source document

[2025-12-22 00:38] lisamegawatts: paper that kinda contradicts me and is also compelling, https://arxiv.org/abs/2501.00663 [2025-12-22 01:24] traves_theberge: Yeah, I've been thinking about this for some time. I've been thinking about this as a model. Data being raw facts "context" Information being that data processed through (n) function Knowledge being the experience / history of what and when it worked. and wisdom being when to take action by memory of (use the context + through (n) function + backed by a history ledger of insights "when did it work") or somthing like that, its not 100% but i'm mulling it over <@1211062099137265723> (files: content.png) [2025-12-22 01:27] traves_theberge: DIKW (Data → Information → Knowledge → Wisdom) describes: How meaning emerges How understanding is layered How judgment and action become possible It answers: What is happening? What does it mean? What should we do? Why is this the right thing to do? [2025-12-22 01:29] traves_theberge: Titans tops out at knowledge, and from my experience doesn't work 🤫 Because it lack a governance and policy / value system.. Wisdom requires governance, not just recall. [2025-12-22 01:41] lisamegawatts: hmm also knowledge to me does mply history, i know it works because x. personally i have a large area in the middle that is stuff i think should work based similarities to xyz but no first hand experience yet if it will work, a theory for understanding why something worked, and when it fails revises why not too, based off of stuff that happened plus data plus inference from priors [2025-12-22 01:44] traves_theberge: of course this is a very immature description but it models the human experience closer than the titian model. going to put my research hat on and work on a architecture / model for this. [2025-12-22 01:55] traves_theberge: (files: content.png) [2025-12-23 08:24] lisamegawatts: https://x.com/omarsar0/status/2000225781927325863?s=46&t=OS71yaTltL7EclsAy81ZcQ [2025-12-26 06:47] optionalsecurity: https://arxiv.org/pdf/2512.09742 [2025-12-26 07:02] optionalsecurity: https://x.com/i/status/2004049715420082585 [2025-12-27 06:33] traves_theberge: https://arxiv.org/abs/2512.07796 [2025-12-31 00:47] lisamegawatts: https://x.com/omarsar0/status/2005999079265034729?s=20 [2026-01-01 08:35] traves_theberge: https://arxiv.org/pdf/2512.22106 check out this paper <@1211062099137265723> [2026-01-01 08:35] traves_theberge: No. I hate this paper lol 😆 [2026-01-01 08:36] traves_theberge: More = More lol 😂 [2026-01-01 08:46] ajaxdavis: tool selection is definitely a hard problem, but hard problems are fun [2026-01-14 11:29] lisamegawatts: https://x.com/rohanpaul_ai/status/2010549041688436828?s=20 [2026-01-14 11:30] lisamegawatts: https://x.com/rohanpaul_ai/status/2011238333142483334?s=20