On Giving Up Understanding

Published on Last edited on

AI has made me question many of my prior assumptions and concepts - what intelligence is, how it can be achieved, its relationship to language. Hence, I was also tempted to question and revisit my relationship to understanding. You see, my current (and prior) perspective on understanding puts it as one of the most important, high-value activities undertaken by humans. It is through understanding that learning can happen. And it is through learning that transformation can happen. This is how we improve - it allows for explicit runtime code modification that produces (more of, better of) a desired outcome! It underpins both creative and analytical tasks. It is, to me, a core part of being human.

But it seems like computers are able to arrive at similar conclusions to those of a person wielding their expertise at a similar task. Which is remarkably cool! Putting aside how grandiose current progress slope looks like, a more pragmatic question is - should we defer understanding to machines?

My answer is a distinctive no.

The same way we should not trade nor give up a real and healthy social life for one that's digital and intermediated by big corps with a different incentive structure, the same level of carefulness should apply to our cognition.

  1. Personal - I don't want to relinquish and give up on something that is intrinsic human, and makes up a significant portion of the joy I get out of working.

  2. Intellectual - Outsourcing the key learning mechanism to a machine would make my brain atrophy (at a faster rate), as I would naturally stop learning⁠12.

  3. Moral - I would not be able to honestly produce any assertions or establish any level of responsibility over my work⁠3, because the machine owns the understanding.

If they sound a bit preposterous or far-fetched, you are also probably right. The conclusion came out of me trying to project far out - >5y horizon, which forces me to assume some model of progress, and this was done over a worst case scenario.

Similarly, how does a good diet look like? Fundamentally, it should be additive to the human being. It should make you better - your brain, your relationships, your work, your outputs.

[1] Kosmyna et al., "Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task" (MIT Media Lab, 2025). https://arxiv.org/abs/2506.08872
[2] Lisanne Bainbridge, "Ironies of Automation," Automatica 19, no. 6 (1983): 775–779. https://ckrybus.com/static/papers/Bainbridge_1983_Automatica.pdf
[3] Madeleine Clare Elish, "Moral Crumple Zones: Cautionary Tales in Human-Robot Interaction," Engaging Science, Technology, and Society 5 (2019): 40–60. https://estsjournal.org/index.php/ests/article/view/260