18 chunks · 16 episodes
...A: Reflective Prompt Evolution Can Outperform Reinforcement Learning. There's a massive amount of capital and effort going into improving frontier models. There's less going into improving the structures we use to extract value from models.
..., aligned just so. When LLMs are tamed with the right substrate, it will create massive amounts of power.
...t so badly that they're chewing through the wires to get access to it. It shows massive amounts of pent up demand. Clawdbot is not the future, but it definitely points the way towards the future. A normie's review: "Clawdbot is directionally co...
...cacophonous media landscape. The entity that makes that curatorial decision has massive amounts of power to shape our world view. The entities who rank the content that is shown care only about "number go up." The result is this toxic stew.
...he raw signal is never seen, it's what is curating your view. The curator has a massive amount of leverage to steer you.[gm][gn][go]
...ill become possible."[ka][kb] But fitting things into an ontology up front is a massive amount of work, and the benefit is only theoretical and indirect. So the direct cost beat the indirect benefit and made it so no one ever did it. But now LL...
...ns in the system and using them to drive an emergent sort. This only works with massive amounts of data, which means only the biggest platforms can do it. But LLMs allow human-level judgment available to anyone, and might make ranking with much...
...ny individual human could do. It does that by extracting the consistent bias in massive amounts of noisy signal flowing through it; the cacophonous actions and decisions of a wide swath of humanity. But it requires a constant flow of these huma...
...n't need strong incentives at all. Writing an app today is a massive slog, with massive amounts of effort before you having anything to show for it at all. LLMs can make programming in the small feel less like a slog and more like being a wizar...
... large and has tons of momentum, and if the plan had to change it would require massive amounts of re-coordination work to update the plan? In that case, "fudging" the measurement a bit doesn't seem quite so bad. Maybe the measurement was wrong...
...he cathedral is a massive, beautiful object with a coherent vision. It requires massive amounts of time, capital, and coordination. Cathedral is top down. It is a high-fidelity manifestation of a particular author's differentiated vision. The b...
...ted in large organizations. Both the cathedral and bazaar approaches can create massive amounts of value. A true cathedral approach is extremely rare and hard to execute. If any point doesn't execute properly, or turns out to be non-viable, the...
...ment, and do the high friction steps to install and onboard. That would require massive amounts of marketing to accomplish… more marketing than the value of the app supports. But that's a limitation of today's laws of physics, not a lack of imp...
...here are a ton of possibly-useful protocols that don't exist, but could... with massive amounts of effort. But LLMs for the first time can work roughly on the clock cycle of computers (or at the very least, a new one can be spun up at any time ...
...absorbing vibes. Humans absorb vibes from our experience. LLMs absorb them from massive amounts of training data. As Gordon notes, not artificial intelligence, but planetary-scale artificial intuition.
...accelerates as you go up the ladder. At the beginning, it has to search through massive amounts of state space to find viable patterns. But with each self-hoisting event, the state space gets smaller and more constrained... which makes it way e...
...er of a thing and its optical size don't necessarily correlate. A tsunami has a massive amount of momentum that can wreak destruction... and yet might not be more than 5 feet tall. "Oh we can ignore that, that wave is only like 5 feet tall." Wh...
...bottom-up sense of systems. The best of them can do a kind of alchemy to create massive amounts of indirect value out of totally uninspiring inputs. The downsides to this playbook: 1) anyone who is watching them at any single point in time will...