{"blyg":"0.4","id":"6f76aft583y28zxgt0h2e0tdbv","kind":"thread","origin":"https://blyg.pioneeringspirit.xyz/","page":"t/6f76aft583y28zxgt0h2e0tdbv/","author":{"name":"PAtwater","url":"https://blyg.pioneeringspirit.xyz"},"created":"2026-11-15T17:33:17Z","updated":"2026-11-15T18:54:31Z","version":1,"content_md":"![[7gpkjct7vr6dq5q2yjkwvj9zq0]]\n\nI wonder if human review will be less important as a new equilibrium is reached? There can be other tests of whether it achieves the goal in specific sandboxes than having an expert go through it. In computer vision and OG ML, there is always the question of what is ground truth. It’s not always clear that humans have the answer there when say counting fish or fields at a scale that dwarf human capacities. \n\nIn the legal domain, I was chatting with a fellow alumni from undergrad. He's now a lawyer and left firm life. His view was that AI enabled more solo practioners and folks with niche speciality skills to thrive. I can see a bar-bell type of dynamic with small and large studios thriving, the latter leveraging in house knowledge to improve bespoke AIs. Some of that involves reviewing. Though I wonder if more of the value will go to explaining and implementing the insighting in collaboration with a client. \n\nThere is a lot beyond just getting something correct symbolically. The map is not the territory it describes, even if it compiles or is clean in the review or whatever. \n","content_html":"<blockquote class=\"blyg-transclusion\" data-blyg-id=\"7gpkjct7vr6dq5q2yjkwvj9zq0\" data-blyg-version=\"1\" data-blyg-origin=\"https://www.msweet.net/notes/\">\n<p>Hmmmm.</p>\n<blockquote><p>The world moves especially fast these days, but the goal of these resolutions is to embrace our important human role in harnessing AI safely, responsibly, in a way that creates value for everyone, and that will be relevant for years to come.</p></blockquote>\n<p class=\"source\"><a href=\"https://slopmanifesto.com/\">No AI Slop: A Manifesto ↗</a></p>\n</blockquote>\n<p>I wonder if human review will be less important as a new equilibrium is reached? There can be other tests of whether it achieves the goal in specific sandboxes than having an expert go through it. In computer vision and OG ML, there is always the question of what is ground truth. It’s not always clear that humans have the answer there when say counting fish or fields at a scale that dwarf human capacities.</p>\n<p>In the legal domain, I was chatting with a fellow alumni from undergrad. He's now a lawyer and left firm life. His view was that AI enabled more solo practioners and folks with niche speciality skills to thrive. I can see a bar-bell type of dynamic with small and large studios thriving, the latter leveraging in house knowledge to improve bespoke AIs. Some of that involves reviewing. Though I wonder if more of the value will go to explaining and implementing the insighting in collaboration with a client.</p>\n<p>There is a lot beyond just getting something correct symbolically. The map is not the territory it describes, even if it compiles or is clean in the review or whatever.</p>\n","content_hash":"sha256:45dd0546ff7bfeab51225abca84c76e9e3a90f1460ace712d61ff4f7a8319aff","media":[],"transclusions":[{"id":"7gpkjct7vr6dq5q2yjkwvj9zq0","version":1,"origin":"https://www.msweet.net/notes/","cited":{"source":"Notes | Matthew McDowell-Sweet","author":"Matthew McDowell-Sweet","excerpt":"Hmmmm. The world moves especially fast these days, but the goal of these resolut…","url":"https://www.msweet.net/notes/752-rust-crabs","retrieved":"2026-11-15T18:54:31Z"}}],"stub_of":{"origin":"https://www.msweet.net/notes/","id":"7gpkjct7vr6dq5q2yjkwvj9zq0","version":1,"cited":{"source":"Notes | Matthew McDowell-Sweet","author":"Matthew McDowell-Sweet","excerpt":"Hmmmm. The world moves especially fast these days, but the goal of these resolut…","url":"https://www.msweet.net/notes/752-rust-crabs","retrieved":"2026-11-15T18:54:31Z"}},"changelog":[{"version":1,"at":"2026-11-15T18:54:31Z","note":null}]}