Compressing reality

As generative text AI goes mainstream on most digital surfaces you might use, it is worth stepping back, looking at how it is used, and learning what can be learned.

One use case is search: looking for information, which was the main use case when ChatGPT came out in 2022.
The other, deployed soon after, is summarizing a long text or expanding a short one into a longer text.

I think the latter is interesting because we generally prompt AI from human inputs: a long essay to summarize, or a few words you drafted that get expanded into a few sentences for another human.

It begs the question of why the essay was long or why the response needed to be longer: culture. Each culture (whether professional or personal) has its own way of being polite, precise, or complete when you express yourself.

One can now wonder how we can still be uniquely human while expressing an opinion, given omnipresent AI usage (and the expectation that it will be used even more in the future).

I think we can draw a parallel to the continuous improvement of compression algorithms invented when the internet scaled to civilization-wide infrastructure: it is worth thinking about who the winners and losers were to get a glimpse of our future.

Coming back to expressing yourself, I would argue that one of the best expressions of an opinion in the post-AI world is to compress reality into the minimum set of words.

In a world where token efficiency will be paramount to business, being able to communicate in fewer tokens (i.e. compressing meaning) that another person can understand because of shared context (i.e. decompressing it) will be a rare and extremely valuable skill.