How I use AI as inefficiently as possible

If you want to be like me and use AI to its least-full capacity, follow these Do’s and Don’ts.

Do give it a vague prompt because you don't really know what you want.

Don't use an AI code editor. Let's face it, if it writes too much code at one time, you're not going to read it. Like a long pull request that you just approve with "🚢 LGTM"

Do ask it to generate names for you. There's nothing better than a company valued at $60 billion coming up with names for a pottery Instagram account like Claylyfe, WheelTurns, MuddyMoments. Actually, I kinda like these...

Don't give AI more context to give better answers. Remember if it's in your head, they can't train off that data.

Do use AI as a glorified Google search. But also use Google to check if someone else has written about what you're searching for.

Don't ask AI to write things in your voice / tone. You don't realise what you sound like until you read what AI writes back to you. Like seriously do I use like that many emojis 😱🤯?

Do use AI as a collaborative partner. But treat it like a partner you don't trust, who makes shit up, and has no idea what you're talking about.

Don't avoid AI. Just be inefficient. Because there's nothing more human than using a tool so poorly you're forced to do things yourself

In all seriousness, I want to start using AI how Amelia Wattenberger suggests in her post, LLMs are a tool for thought.

Not as a tool to solve my problems. But as a way to gradually reach the solution myself. Inception style.

Instead of asking it to generate code, have it validate my approach. Or instead of asking it to edit my posts, ask if there’s gaps in my thesis. Ultimately, I would still do the work and get the learning goodness. But AI can help me get there faster.

At this point, I’m just summarising Amelia’s post. Go read it if you haven’t already! It’s beautiful.

See you next week!