The dry Italian landscape rushes by outside the train window — hills, small houses, tunnels, and power lines. We are traveling from Rome to Lake Garda. Over the last few days, we have explored the capital and all its sights: the vast museums of the Vatican, the hidden squares in the old town, and the burning sun in the Roman Forum.

I’m a fairly normal person, living a fairly normal life, doing fairly normal work. And, like many others, I think a lot about AI — how it is working and how it might potentially change our world. I am in no way an AI propagator selling the promise of effortless work, nor a doomsday skeptic who only sees problems with it. This post is in no way an attempt to sway anyone in any direction. It is merely a way for me to process some recent thoughts on the topic, and I expect to change my footing many times before we know how this ends.

This is a bit different post compared to what I usually write about. But, hey, this is my blog, my rules. I write about whatever I want… The last few days I visited the Sweden Rock Festival, without a doubt Sweden’s biggest rock and metal-oriented music festival. This year there were more than 90 bands in the lineup and 40000 ticket holders according to Swedish Television. A lot of people and a lot of music.

LLMs and AI are what everyone is talking about these days. Tiring, I know. But it’s an important topic to discuss — whether we want it or not, it will change our lives and how we live. Better to do our best to share our opinions as much as possible. At least it might change something or someone.

First off, I’m an AI user, and I pay for a subscription to the service I use. I find it helpful as an accelerator when doing certain tasks, such as preparing classes in the form of practice exercises, exams, and outlines. When coding, it helps me speed up troubleshooting and acts as a sparring partner. I never use AI to generate things from scratch. There are simply too many problems with the output to use it even closely reliably for that purpose.

Yet another reflection based on a session at myConf. Erik Dörnenburg talked about GreenOps — ways to utilize your computing resources as efficiently as possible in terms of environmental impact. It was a quite interesting talk about moving your resource needs to different locations in the world depending on where green electricity is available.

There were also several other considerations for lowering the carbon footprint, such as using cloud computing. Cloud servers can run at higher loads and are therefore more optimally utilized. Another point raised was the effectiveness of different programming languages in terms of power consumption. During the presentation, data was presented from a well-known research paper titled Ranking programming languages by energy efficiency. Essentially, it finds that compiled languages are more energy-efficient than scripting languages. But, as always, there are opposing views — one of which is outlined in It’s Not Easy Being Green: On the Energy Efficiency of Programming Languages.