Delegating to AI

If you amplify 0, you'll still get 0.

When people delegate their work to AI, it's quite similar to managers doing the same with their team.

But delegation is a skill, and when it involves hardcore technical elements, it is even more difficult. Moreover, you need to review the output at some points, since eventually managers will be accountable for the results.

Real managers know when to go hands-on, even though they might not know the details of the implementation, especially when things go technical (which always do at some point).

Prompt engineering is a form of delegation. You need to ask in a certain way, in a certain sequence for work to be executed. There's no one universal way to ask for things to get the work done. It usually comes with a process, a feedback loop, and eventually nailing down the task or the work to be done.

I see one-man armies already doing the work of a team; therefore, they will be able to do the work of a whole company with AI.

But that requires a stack of skills which, in return, requires understanding - but more importantly, knowing how to manage some fundamental concepts.

If you don't have those fundamental skills, then you actually don't know what to ask or, that could be worse, you might get something but have no idea if it's gonna be relevant or not for your work.

Overnight, we were taken off-guard by having services that provide personal assistants and a team ready to work for us for a dime (I would even say for free.).

Conclusion: Steve was right; to leverage this power, we need to stay hungry - that is, connect the dots in a way we would never have thought before.