1,598 mentions · 615 chunks · 117 episodes
...tural schelling points requires cognitive labor that used to be scarce. But now LLMs might help sift through and find the obvious schelling points. Of course, this would change the internal politics meta-game…
...ently. It used to be a pain to think through it for each decision. But now with LLMs and their infinite patience, it's easier to have a fuzzy set of values and mission statement operationalized. Every decision can be run through an LL...
LLMs are losing the ability to simulate real people. LLMs are largely a warped mirror of all of the human input. It used to be possible to use LLMs as a k...
Apps are the wrong soil for the seeds of LLMs to blossom into their full potential.
Three pace layers of prototyping with LLMs: 1) The LLM does everything. Expensive, loosey-goosey, flexible 2) The LLM behavior is sublimated into a mechanistic harness that can be run inside o...
... but go to a different source you'll get a different answer. The monoculture of LLMs leads to everyone having the same answers to the same questions.[ck] I heard of someone who thought someone else had plagiarized their essay. Turns o...
This week's round up of "we're in the wild west era with LLMs": A postmortem for a vibecoded tool called DrawAFish that had abuse problems. A Cursor exploit that allows arbitrary remote code execution. AgentFlay...
The most important ability of LLMs is infinite patience. What kinds of information tasks could normal humans do if they never got bored?[cy]
...context as possible, you want the right context. The wrong context confuses the LLMs and makes them spiral out of control, losing the plot. What you want is the smallest amount of context that will give the LLM what it needs to give y...
Imagine a system where LLMs generate new combinations that a community of humans curate emergently through their individual authentic actions. LLMs help create "patterns": littl...
...ever took off outside of niche applications. Maybe that changes in the world of LLMs?
...y code. If you're a curated programmer you'll get a lot of curated code. Before LLMs, sloppy programmers could at least make a lot of progress, which was an advantage. But now the LLMs do that part for free and for everyone. The balan...
Don't let LLMs make load-bearing security decisions.
Anthea's newest piece compares LLMs as freeing a caged tiger. The thinker is a caged tiger set free with intellectual collaborators willing to go wherever you want to go. A cage is also...
...people making real decisions that align with their authentic needs and context. LLMs are a fossilized version of real people's decisions; it can't pick something novel. It looks superficially the same, but it's fundamentally different...
...atabase to a real consumer need requires lots of complex specialized stuff. But LLMs might not need that. It's kind of weird that the LLM model creators also have consumer frontends to them. It shows how powerful LLMs are that they ca...
...en the actual LLM-native software yet. The software that takes for granted that LLMs exist, not as the primary input, but as a secondary one.
Code is cheap to build, expensive to have. This is even more true in a world of LLMs. YAGNI.
There is no solution to prompt injection in systems where LLMs call the shots. LLMs seeing raw data and being asked to make load-bearing security decisions cannot be made safe, no matter how good the model gets. ...
...ld add the word "useful" in front of intelligence. You could imagine having two LLMs that require huge amounts of compute locked in an infinite debate spiral about how many angels can dance on the head of a pin. Or more likely: a red ...