The Promise of A.I.

Lately you can’t swing a cat without hitting an article or a podcast about how AI is going to change everything. Easy, cat lovers – it’s just an expression – nobody’s swinging any cats. My impression so far has been mixed. I’ve seen it do an amazing landscape proposal based on a xeriscape philosophy and the available space. It does a decent job writing a fugue based on notes you feed it. It can help you allocate your 401K. It can take a bland recipe and spice it up. Then write a Haiku about it.

I have also seen it look at engine data and come up with a perfectly plausible explanation for an event that’s a complete crock of shit. The problem is it either doesn’t know it’s full of shit (the Dunning-Kreuger effect) or it doesn’t care. The problem is you give it another set of data and it nails it. You just can’t tell one from the other without doing the work yourself.

Pilots of single-engine piston airplanes are trained from day one to look for a suitable landing spot if the engine fails. It’s not unusual for me to begin flight planning with a direct route then rubber-band it to fly over a nearby airport. A few years ago Foreflight added a glide ring to its software. The ring calculates altitude and wind direction and speed and tells you if you can glide to that nearby airport or not. In calm wind it’s pretty much a circle – in heavy wind it’s more cardioid shaped. It’s neat to see it moving along and change shape as you fly.

Foreflight just took it a step further and now there’s a button. Press it and you get a list of airports sorted by runway length and wind orientation. So airport A might be closer but airport B is downwind and has more or longer runways. So it’s like a really good copilot helping you process a lot of available information to make better decisions at a crucial time.

That’s going to save somebody’s life someday. Now we’re talking.