People (even from the same culture, with the same native tongue) use words differently. It is one of the reason for which I am mildly skeptical about all such t

People (even from the same culture, with the same native tongue) use words differently. It is one of the reason for which I am mildly skeptical about all such things as "How happy you are on the scale 1-10?" - no matter what is the sample size (it could be a billion), it depends on one's inner definition of "happy" and on the numbers attached to it. Words are virtual. At the same, experiences are real. Also, real is their impact of our lives. "A headache" can be anything for from a slight discomfort, to something crushing, putting in a blank stare. It can be of various lengths, various frequencies, impacting us in various ways. Both matter. > I want practitioners and researchers to help us name things so that they can measure them, and so that those measurements can be accurate reflections of our internal realities. I think it is one of things where LLMs can shine - by translating one's own notes, ones own recordings of everyday life, into a summary. This used to be a bottleneck - so people HAD to resort to some questionaries (binary or Likert scale), to make it quantitative. Now we can automatically process actual text. (It was possible before, to some extend, with simpler models like word2vec or Latent Dirichlet Allocation), but with LLMs we can actually get the full context, not only look for keywords, their groupings and associations.