AI: Getting Results

Development

This will all likely be obsolete by the time I’ve even finished writing it, but here’s a couple of things I’ve noticed as I work with these technologies:

Be Specific, but not too specific

– Try to balance specificity with overall intent

– Consider design principles over specific workflows

– Guide the model to “think” about what you’re trying to achieve

– Provide examples and screenshots when possible

Create a plan

Like any project, planning is key. Make sure you work out a solid plan, working out what technologies you’ll be using, the frameworks you need, etc. I’d use the most clever bot at your disposal to help you formulate and refine your plan before starting a project with your coding assistant of choice.

Watch it write and pay attention

These things can occasionally go nuts on you or start working in endless loops. Sometimes even the brightest bots will miss simple things like closing tags and semi-colons. It’s best to pay attention to what they get up to, even if most of the time, it’s fine.

Review all code before you ship.

You’re likely doing this already, but go ahead and give everything a once over. I’ve even been asking the bot to review itself, which often fixes issues it had created earlier. These tools can sometimes be too linear in their thinking, so they break things during revisions.

Illustration

This is going to be its own whole thing. I’ll write out a few tutorials on how I use Invoke, as well as a few samples and link them here.

AI Overview

Getting Results

Ethical Concerns

Policies

Third logo test