When working with AI, humans are the bottleneck limiting efficiency.
I’d like to share something I’ve learned.
Because AI is so efficient, my habit is to start with one main task, give it clear enough context, and let it get going. The first few rounds usually go quite well.
But while I’m waiting, I want to make use of the time, so I start another thread to work on a different task.
At first, that’s fine.
The problem usually comes after both tasks have been running for a while. One of them will reach a point where you need to truly understand the context and give clear instructions before it can continue. But by then, my mind is actually focused on the other project—or, worse, on Twitter, a newsletter, or some other distraction I turned to out of boredom while waiting for AI to work.
At that point, it’s hard to bring my attention back to the context. It’s like an elephant that’s spotted a banana and wandered into a banana grove: once it’s there, it’s almost impossible to steer it back onto its original path.
So I start giving AI too little context and letting it make its own judgment. That’s when the quality of its output starts to decline. Or I step away from the computer and the work simply stops. When I pick it up again later, I can’t just resume where I left off; I have to rebuild the context from scratch, and naturally, efficiency takes a hit.
So here’s what I’ve concluded: it’s fine to have two projects going at once, but one of them must be the main focus. If you have a side project, it’s best to choose something simple that doesn’t require much context or judgment and mostly just needs confirmation. Otherwise, you’ll quickly run out of mental energy, the project quality will suffer, and the worst part is, you’ll know you haven’t done a good job.
I’d like to share something I’ve learned.
Because AI is so efficient, my habit is to start with one main task, give it clear enough context, and let it get going. The first few rounds usually go quite well.
But while I’m waiting, I want to make use of the time, so I start another thread to work on a different task.
At first, that’s fine.
The problem usually comes after both tasks have been running for a while. One of them will reach a point where you need to truly understand the context and give clear instructions before it can continue. But by then, my mind is actually focused on the other project—or, worse, on Twitter, a newsletter, or some other distraction I turned to out of boredom while waiting for AI to work.
At that point, it’s hard to bring my attention back to the context. It’s like an elephant that’s spotted a banana and wandered into a banana grove: once it’s there, it’s almost impossible to steer it back onto its original path.
So I start giving AI too little context and letting it make its own judgment. That’s when the quality of its output starts to decline. Or I step away from the computer and the work simply stops. When I pick it up again later, I can’t just resume where I left off; I have to rebuild the context from scratch, and naturally, efficiency takes a hit.
So here’s what I’ve concluded: it’s fine to have two projects going at once, but one of them must be the main focus. If you have a side project, it’s best to choose something simple that doesn’t require much context or judgment and mostly just needs confirmation. Otherwise, you’ll quickly run out of mental energy, the project quality will suffer, and the worst part is, you’ll know you haven’t done a good job.