I spoke with the CEO of a FinTech company who really wants his org to start using AI so they don't get left behind. Fair enough. Then I spoke with one of his analytics managers whose team recently started using ChatGPT. They got some benefit, but they're doing it completely wrong.

They dump files into the chat interface without knowing which model to choose or what mode to operate in. They're getting hallucination problems because nobody explained the difference between a reasoning model and a non-reasoning model, or agentic mode vs non-agentic mode. They just... threw it at the wall.

This is happening everywhere.

The FOMO-FOBO Sandwich

Most AI implementations I'm seeing fail for the same reason. You've got management in full FOMO mode who get sold on some AI stack like Copilot and then engineering has to adopt it like a child that they need to nurture.

On the other side, engineers are dealing with their own crisis. FOBO: Fear of Being Obsolete. Some of them are resisting the change by half-assing it. They give poorly defined prompts, get garbage output, and then say "see, AI sucks." They're not wrong that it sucks sometimes. But they're also not trying.

The Options Paralysis

Here's another thing nobody talks about: the sheer confusion around how to even use this stuff.

There's the chat interface (which itself has multiple models). There are coding tools. There's the API platform. There's terminal development now. There are plugins in Excel, PowerPoint, browsers. There are AI browsers and there desktop apps that does multiple things. It's overwhelming for anyone trying to figure out where to start.

The non-technical folks don't know the difference between these options but they know they need AI. The technical folks who aren't business-oriented are resisting, which creates negative feedback for management. And management just sees that their AI initiative isn't working.

What I Actually Think

There's definitely a lot of noise and fluff out there. But this technology is also very real. It will redefine how companies operate and how work gets done.

I see an opportunity for technical people who have business sensitivity and are willing to learn what's actually possible (and what's not) with this tech. If you can bridge that gap, you can create real impact within your org or help other orgs get there.

Short term though? There will be blood. Manual testing, customer support, front end engineering, bureaucratic dev work like writing parsers and updating integration APIs. These are going to get hit hard. And as AI gets smarter, the impact will creep up the stack to higher level roles too.

I'm not sure what the long term workforce impact is going to be. Will it increase jobs or reduce them? My take is that a lot of time is going to get freed up. That time can either go toward producing more or doing higher level strategic thinking.