13 chunks · 8 episodes
... your guesses will not align with their assumptions. Your mental models are the frog DNA that fills in the implicit assumptions you left unsaid. Some mental models are obvious and widely shared; some mental models are specific to your exp...
Anthropic Artifacts is 100% frog DNA. It can whip up a little interactive thing for you based on an English language prompt. But all it has to work with is what it absorbed during traini...
...lub. The "answer" could be a pop culture reference everyone already knows (like frog DNA from Jurassic Park) or a simple evocative metaphor (doorbell in the jungle).
LLMs fill in underspecified parts of the users' request with frog DNA. The frog DNA is inherently mushy; average. This means that the under-specified parts become more average, pulling toward mediocrity. That's bad… but...
...ences have some inherent squishness. They fill in the underspecified parts with frog DNA, reasonable guesses. Even with the fully specified parts, sometimes they just… forget. LLMs are not deterministic. Well, technically if given the pre...
...it as a thinking partner. LLMs' ideas are never good. They're always mush, just frog DNA. I was chatting with an author who told me he refuses to use an LLM. He told me you "write what you read", and he feared that the more he's exposed t...
LLM-generated software is mush. It's 100% frog DNA. The LLM extrudes out a hyper-generic answer to your specific query on demand. But what if there was someone else who in the past had done precisely ...
...l in the gaps? If you have to fill in the gaps from nothing, you get 100% mush, frog DNA. But if the user has given a bit of a hint of the type of data, or their intent, often an LLM can fill in the gaps in a way that is likely what the u...
A RAG approach is kind of like Shark DNA vs Frog DNA. 'Frog DNA' is the generic mush the model learned, the background context it falls back on to fill in the gaps that you didn't specify in your prompt...
Imagine a spec expanding into code. The LLM uses frog DNA to fill in any ambiguities. But as the model improves, the frog DNA gets higher quality, and the outcome gets better. The quality of the output depen...
We use "frog DNA" to fill in gaps. When confronted with ambiguity, we use our pre-existing knowledge of the world to guess at the resolution to the ambiguity. Like in...
Frog DNA is average, mush, generic. That's one of the reasons LLMs pull everything to the bland centroid; they fill any ambiguities with mush. What if you cou...
...n the DNA, so you need something to fill the gaps. You use a new baseline, like frog DNA, to fill in the missing parts. When you retrieve the memory, your baseline understanding may have evolved. So you fill in the gaps in the memory with...