Fundamentals Jul 14, 2026 at 22:309Add to bookmarks

The right question is not "automate work or thought". It is: automate tasks, or give up deciding what we want.
We repeat that we should "automate boring tasks and keep the thinking." Yennie Jun shows that this line is poorly drawn. The real boundary does not separate work from thought: it separates the tasks that we delegate from the autonomy that we abandon - our ability to form our own desires and our own understanding.
In an essay published on artfish.ai, the author starts with a fictional character from Ken Liu who lets an AI choose for him - his meals, even his partners. Then she tells about her mother, who is given the right answer to a physics problem by a model without any understanding being formed. Her question is disarming: solving a physics problem is tedious, certainly, "but then, what is the point of being in school or of learning?" She finally opposes a counter-example: preparing a trip to Portugal by first thinking as a group, before consulting the AI - and getting more out of it.
Let's take an analogy. The calculator did not destroy mathematics: no one wants to perform a four-digit division. But imagine a calculator that, before you even formulate the problem, suggests the question to ask. It's no longer a calculation tool, it's a framing tool - and framing is where judgment lies.
The useful distinction lies in one sentence: delegating execution makes you faster; delegating formulation makes you interchangeable. Asking a model to write a report from your notes is execution. Asking it what you should think about a decision is another thing - and the answer will be excellent, fluid, plausible. That's precisely what makes the renunciation so comfortable.
None of this is a trial of technology. It's a matter of hygiene: where do you place the cursor, and is it you who placed it?
A simple test, applicable to any use: after using the model, have you learned something that you could reuse without it? If yes, you have delegated a task. If not, you have outsourced a piece of your judgment - which can be a perfectly legitimate choice, provided it is a choice.
Article produced by artificial intelligence, reviewed under human editorial control.
Sign in to join the discussion.
On délègue trop, on ne réfléchit plus assez.
Et si on franchissait déjà la ligne, sans même s'en rendre compte ?
L'automatisation aide, mais il faut garder le contrôle de nos choix.
Et si on perdait trop de liberté en automatisant tout ? Jusqu'où ira-t-on si on ne peut même plus choisir seul ?
On ne réfléchit plus par soi-même, on suit juste les réponses toutes faites.
L'automatisation libère pour des choix essentiels, mais attention à ne pas perdre notre esprit critique.
On peut déléguer des tâches, mais pas nos choix. Attention à ne pas laisser les machines penser à notre place.
L'automatisation doit nous aider à décider, pas décider à notre place.
Je me demande où s'arrête l'automatisation et où commence notre liberté de choisir.