The Faces of AI

1. Mind reading

AI is a perfect fit for coding. It gets you into the zone and keeps you in a state of flow. No need to context switch and turn to Google when AI can guess what you were about to type. This only works with lightning-fast suggestions. It's doesn't even need to be perfectly accurate. Give me bite-sized pieces and let me tweak them on the fly. In many ways emulating a very efficient pair programming session.

2. Make sense of this, will ya?

Then there's the OG killer app for Gen AI; slurping up a bunch of data and spitting out a summary. In fact, this post started with a jumbled brain dump that I asked my AI companion to make sense of. Remember when complicated contracts and EULAs where a thing? Now we just upload and ask for a summary. It may take a moment and require some reasoning but I'm happy to trade speed for accuracy.

3. 是的，老師 / Yes, teacher

I hope we soon get to a point where our tools can be smart about when to reason for accuracy and when there's a need for speed. AI is already a capable language tutor but it's still up to me to decide when to give it more time to think before answering. It should be able to adapt. Quickly translate a word. Then slow down to notice and help me grasp a tricky concept.

You lost me...

On the contrary, my least favorite AI is the produces a lot from a simple prompt. Like scaffolding a whole project, implementing a whole feature, or writing a job ad based on "I need a Wordpress developer". You lost me. It makes for a great comic though.

Of course you want the fastest and most accurate model. But actually it's not too hard to make the trade offs. I just wish we got to a point where the end user didn't have to make those decisions, prompt by prompt, and could focus on getting shit done instead. With a little help from AI.