79 chunks · 60 episodes
...and criticism. (Apparently this is an idea from David Deutsch.) That is, form a mental model, a hypothesis. Then expose that hypothesis to disconfirming evidence, for example ground truthing it. Guess and check. The disconfirming evidence giv...
...box[wg]" frame for LLMs quickly leads to icky scenarios and also gives a flawed mental model for what they can do anyway.
To guess what someone means in an ambiguous situation, you need to overlap on mental models. If you don't share mental models, your guesses will not align with their assumptions. Your mental models are the frog DNA that fills in the implici...
...ions with humans.[yr] Another way that the "LLMs are basically a virtual human" mental model is wrong. LLMs have all of the background world knowledge that was statically baked into them during training, but their only "working knowledge" is ...
...ss quality that it can be compelling on its own. But they also cement the wrong mental model: "LLMs are just like a person, but virtual." That vastly misunderstands what LLMs are, what they can do, how they could be used. LLMs are an alien br...
... early adopters. The early adopters can pull you in weird, random directions. A mental model to get the underlying dynamic: Imagine an underlying distribution of points with 0.7 correlation. Imagine adding one point from the distribution at a...
A mental model for retconning your thing: what is a plausible story that is compatible with you being a master of the universe. If you were a master of the universe...
...-competent person. That is comforting and easy to use… but also gives the wrong mental model for what they can do. LLMs can perform feats of patience and recall that no human ever could. If you don't think about how LLMs are different from hu...
...ads of my curiosity, in some cases leading to deep insights that reconfigure my mental model of the world or myself.[alz][ama][amb] These conversations[amc][amd] are long and meandering. They have various dead ends or odd paths that I later d...
... then later more signal comes in that requires our brain to snap to a different mental model. Various optical illusions trigger this reliably. When it happens, there's a kind of whooshing vertigo feeling as the whole world reorients around yo...
...ing party trick to show people how open-ended the system is. You break people's mental model of what is possible, and suddenly everything seems possible.
The expected nastiness of a surprise is tied to the fidelity of the user's mental model and the stakes. An inaccurate mental model and high stakes is a recipe for a massive nasty surprise.
...essfully own the system? That is, maintain, fix, and extend it with an accurate mental model? If not, it's not known to be sound. Soundness normally requires careful layering, making sure that each layer is thin and understandable. Thinner la...
...leverage. If not, then the magic could have a very different behavior than your mental model of it. Of course, in practice we can't peel back every layer; that's the whole point of abstraction. But the bar is could you have understood it and ...
I love Anthea Roberts' concept of dragonfly thinking. Any mental model we apply to a problem is a lens. A lens must reduce the signal of the real world into an easier-to-consume distillation. Lenses are great, because th...
...y than the absence. The absence is easy to forget about. You have to maintain a mental model of the now-hidden thing, and keep refreshing that memory or it evaporates from your awareness. The presence is much harder to forget about--it's righ...
... an answer that is superficially great but actually bad for a subtle reason. My mental model is a Simon Giertz-style ketchup robot. After a few minutes of work the LLM agent plays a triumphant chime and happily delivers you… a steaming turd. ...
A mental model: data is "radioactive" if it could be tied to someone's identity. If that data touches other data, or is shown or shared in the wrong context, there ...
A thing that makes it fun to play with your system: users have a rough mental model of "I bet if I did X with Y I'd get Z" and they do it and something interesting happens, even if it's not precisely Z. Especially if there's an undo ...
...ou kick the tires of a problem and it's like you expected. In these cases, your mental model doesn't have to update much, you just reduce uncertainty. Sometimes you kick the tires and it's like kicking over a cardboard cutout and discovering ...