It's not that good.

Maybe (definitely) it's a me problem.

I've been watching way more YouTube (probably than ever) lately.

I've mostly been watching content from Tina Huang on how to 'use' AI: how to write good prompts aka prompt engineering and how to use context to drive better outcomes from your AI friend(s) aka context engineering. Put simply, of course.

But I didn't start here.

As someone that usually reads to learn, I kept finding myself with a bunch of browser tabs open on AI and [insert topic of your choosing here, for me, usually patient behavior, … surprise!). Some of these articles I'd read and had thought Cool! Let me hoard this on my browser. But others (fine, most) I'd half read and many I'd opened because I thought I should read them: they'd been recommended by other content I respect/admire or were... clickbaity. Sue me.

While I tried to read my way to wisdom, I fell firmly in the camp of OK, is AI really that great?

I was (pretty darn lazily) slinging a single sentence or two prompts into Chat GPT and was whelmed (at best) by the output.

So it was easy to roll with the 'AI is just not there yet' crew. So close yet so far.

In many ways, though, it turns out it was really more likely a me problem.

I wish I had some Magical Turning Point that led me to watching YouTube and ultimately learning by doing (see one, do one, teach one... a sentence long repeated to me that apparently has failed to sufficiently cross my mental chasm).

Am I still slamming into posts with no words I can understand. Yes, absolutely. But I've gained some self-efficacy by building an AI Agent that sends me daily updates of [admittedly super dumb stuff].

The adoption and learning curve is super under-appreciated in AI, as far as I can tell. Solo and small operators can implement and see advantages quickly but my (somewhat) speculative belief is that we can't just rely on people to use AI when it's available to them.

Do people use free gym memberships or other perks?

No, not usually.

Will we implement till the cows come home, claim it doesn’t work, and then scale back? Probably not.

But it seems that without the right training wheels, the canyon between the bots and the nots is about to get a lot wider.