Now the AI is Getting Interesting

AI stuff has been pretty boring to me.

Granted, I haven’t made the time to dive deep or spend a lot of time contemplating the full implications and use cases. But the early generated images seemed kinda cool but not “Oh shit!”, and the generated text seemed like more of the same internet-molded crap humans already produce too much of (old man shakes fist at cloud).

AI search has thus far seemed like an even worse version of what Google is becoming – highly constrained and censored, with an express mission of obscuring points of view that don’t fit the comfort zone of the programmers. Boring and kinda creepy.

But then HubSpot launched ChatSpot.

I haven’t used it yet, and have heard mixed results on its performance. But that will come. What got me excited is the replacement of complicated processes of generating reports from a CRM with plain english requests.

If you’ve never worked in a company that uses SalesForce or HubSpot, or god forbid something like Razers Edge or Aptify, you have no idea how hard it can be to surface seemingly simple info that already lives in your database.

“OK everyone, welcome to employee training number 11, where we’re covering how to create multi-conditional lists and then link them to an event type and generate a report”, says the weirdly excited ops person when all you wanted to do is see how many lapsed customers registered for your webinar.

The process involves about seventeen hundred clicks, infinite scrolls through drop-downs, and many conflicting, redundant, and counter-intuitive names for things (tags, categories, labels, fields, entities, etc.) If you flub up on one of the many conditions, the whole report is bogus.

The ability to tell the software the same type of thing you’d ask your ops person is incredible. “Get me a list of every former customer who registered for this event.”

So that’s one use case I love. Truly wealth-creating, in that it allows humans to accomplish more with less.

But there are more.

Ever had to get on the phone to figure out why the cell phone company charged you incorrectly and fix it?

Ever had to navigate a government bureaucracy online or over the phone?

These are basically the same as getting info out of your database, except the people on the other end are grumpier than SalesForce.

Imagine an AI assistant to do all that for you.

“Remove my oldest kid from our cell phone plan and shop around and send him the best individual plan you can find along with the price.”

“Update my home insurance policy now that we put in a pool.”

Another potential version of AI that excites me is one that is not constrained or controlled by its creators (who are rightly fearful of social and political pressure). If you could hone and adjust your AI yourself (without needing to be a programmer) then things like search could actually be useful, instead of Orwellian as they seem so far.

I’m not worried about AI or scared of it. I’m always worried about humans and our potential to do great evil with anything from a rock to a rocket. New tech opens new risk areas for sure. But it opens new opportunities as well.

I’d like to see AI help solve the mundane stuff on a scale that creates serious standard of living improvements. The creative side with AI art and literature doesn’t do it for me so far.