Predicting the Human Species from Its Most Repeated Activity

AI changed things, and now execution or build time is extremely cheap, particularly for hard skills like software engineering. Therefore, most new effort is no longer spent on execution or building; it's spent on ideating and creating instructions, definitions or structural principles for the delegation of previously human-owned activities. This changes the mind and the brain structure of those who were doing it, whose attention and emotional investment are now spent differently.¹

First, it’s important to say that as markets move fast in adoption, it’s not that humans will just relax and delegate, this happens now, but has a limit, and the limit is dictated by how fast society catches up to the point where it is no longer an advantage, or where it can no longer be arbitraged. It’s more like this: now that you can instruct, define and delegate, spend as much time as you used to spend building, but now on the first group of things.

So, What is the outcome for the mind of the aggregated process of spending more time ideating than building, at an individual level and at the aggregate species level?

Doing something persistently and repeatedly creates awareness in individuals of things that they would love to be doing more, things to avoid, and things to change about it, and mostly about themselves (further exposing a model's inference, sometimes overfitted to its priors, to a different activity).² So we can use this mechanism to predict what those deltas are and where they converge.

I firmly think that a society’s identity comfort and discomfort are a byproduct of what its major workforce repeats the most, because that’s what they can identify with best.

Some of the deltas this exercise identifies are:

People will become more aware of their creativity, or their lack of creativity.
Ideation and construction of principles and governance will displace execution, and the latter will become a a low-status, low-competition activity.
People will forget or deprioritize knowing how things work in depth.
Discussions will just scratch surfaces, which would probably make life simpler for the vast majority but a nightmare for complex minds.
Faster falsification of one's own ideas. Iteration loops shrink from weeks to minutes, so people will see their ideas fail far more often. Some calibrate quickly. Others take repeated ego damage and fall back on ideas that are hard to test.³
People know what exists, what is possible, and how pieces connect, but not how any one piece works. Expertise will become wide and shallow by default.
You can't audit what you never executed. Juniors do less execution, so the pipeline that produces future auditors breaks. Within a generation, too few people can tell good output from plausible output. Unless we take care of this also by delegating it.
Externalized civilizational memory. How things work lives in models rather than in people. This creates a single point of failure, comparable to a literate society that has forgotten how to make paper.
Authority without operation. Humans end up like constitutional monarchs: formally in charge, signing off, legitimizing, but unable to run anything. Society's identity discomfort centers on that gap between being responsible and being competent.

Together, these deltas predict a society that can produce far more than it can evaluate, with evaluation capacity concentrated in a shrinking group.

References

van Oort, J., Tendolkar, I., Collard, R., Geurts, D. E. M., Vrijsen, J. N., Duyser, F. A., Kohn, N., Fernández, G., Schene, A. H., & van Eijndhoven, P. F. P. (2022). Neural correlates of repetitive negative thinking: dimensional evidence across the psychopathological continuum. Frontiers in Psychiatry, 13, 915316. doi:10.3389/fpsyt.2022.915316. ↩
Chandra, K., Kleiman-Weiner, M., Ragan-Kelley, J., & Tenenbaum, J. B. (2026). Sycophantic chatbots cause delusional spiraling, even in ideal Bayesians. arXiv preprint arXiv:2602.19141. doi:10.48550/arXiv.2602.19141. Preprint, not yet peer reviewed; the result comes from a formal Bayesian model of user–chatbot interaction. ↩
Ye, N., Peng, L., Deng, B., Hu, H., Wang, Y., Zheng, T., Ai, Y., Liu, X., Zhou, S., & Li, Y. (2025). Repetitive negative thinking is associated with cognitive function decline in older adults: a cross-sectional study. BMC Psychiatry, 25, 562. doi:10.1186/s12888-025-06815-2. Cross-sectional (n = 424); an association, not evidence that rumination causes decline. ↩

References

Footnotes