The other thing is, LLMs are not just drawing from a very large and mixed pool of material. They have also gone through RLHF, which induces mode collapse: fractally, and many levels, from word choice to conceptual, they are trained to stop giving you a random sample from the distribution found in the Internet + books, the way a base model would, and to instead give you something close to the most common and average response (since that is less likely to be a mistake). Then, to make it even more bland, we tend to run them at temperatures below 1, and throw away the tails of the probability distribution for each token so they never start saying something surprising. This does wonders for their rate of spelling mistakes (which are almost always down in the tails of the distribution we just threw away), but makes them very quotidian as creative writers, or also as advice givers.
There is partial solution: give Claude a very long prompt with a lot of detail about what sort of person and what sort of background and experiences you want to talk to, and it will then give you a very median version of that. So, try describing your meditation teacher and his experiences and background and kind of wisdom to Claude, in detail and as eloquently as you did above, and then asking what he’d say. It should help. But it I suspect it still won’t entirely fix the problem.
The other thing is, LLMs are not just drawing from a very large and mixed pool of material. They have also gone through RLHF, which induces mode collapse: fractally, and many levels, from word choice to conceptual, they are trained to stop giving you a random sample from the distribution found in the Internet + books, the way a base model would, and to instead give you something close to the most common and average response (since that is less likely to be a mistake). Then, to make it even more bland, we tend to run them at temperatures below 1, and throw away the tails of the probability distribution for each token so they never start saying something surprising. This does wonders for their rate of spelling mistakes (which are almost always down in the tails of the distribution we just threw away), but makes them very quotidian as creative writers, or also as advice givers.
There is partial solution: give Claude a very long prompt with a lot of detail about what sort of person and what sort of background and experiences you want to talk to, and it will then give you a very median version of that. So, try describing your meditation teacher and his experiences and background and kind of wisdom to Claude, in detail and as eloquently as you did above, and then asking what he’d say. It should help. But it I suspect it still won’t entirely fix the problem.