The benefit assigned to using LLMs, the same one attributed to any automations, is that it saves time. But what do you do with the saved time?
In certain moments, if it comes to this, after I revise a plan and instruct an agent, I might observe the code modifications, but then I switch to my browser and first instinct is to check Twitter… I am not doing any deep work with my fragmented attention. Afterwards I feel sluggish, and worse case scenario, I end up tweaking the AI generated code outputs it gets more daunting than writing it by myself from scratch.
The question of this note is also part of the reason why I cannot get on the utopian bandwagon of a post-work society. There is always something to work on, and do people really think the state will subsidize them? The AI companies? Not to mention the concentration of power and capital that exists.