LLMs?

A few years ago LLMs were at best a fun curiosity. There were many reasons not to like them. They were too unreliable to put much faith in. To many of the tech influencers that I trust, they were a bit of a joke. At my employer, the people embracing them ended up creating more work for others with weak PRs and excessive text in everything they did.

Some of these things are still problems. When people use these tools excessively, they might look strong at easier tasks, but struggle mightily when the hard tasks arise. Historically, the easier tasks would have prepared them for the complexity, but now they appear to be making it harder to help people prepare to tackle the hard stuff.

I dabbled with using ChatGPT or CoPilot here or there, but never went too deep. There was too much noise in the market, too much hype, and too much real work at my job that needed to get done.

Recently things have changed. Those skeptical tech influencers who mocked the state of AI now have agents working for them constantly. At my employer where it had been cool to regularly poke fun at the AI slop showing up, I’ve started to see some of the most skilled and influential developers embrace new ways of working.

I’ve read some interesting articles recently. Was somewhat influenced by the 2025 year in review podcast from Python Bytes Was pushed deeper by a listen to Joe Reis interviewing Wes McKinney And was then pushed over the edge by reading some of Wes’ blog posts and podcast transcripts

I guess it’s time to see what I can do.

For the next month or more, I plan to be intentional about trying to figure out what is really going on, and see what heavy use of these tools actually feels like.

Today, I don’t even have a good grasp on much of the basics. I think I use words without understanding what they actually mean.

Some of the questions I have are:

Let’s see where this takes me.

Hopefully by the end I’ll at least have a better formed opinion.