I learned a couple of things today.
First and foremost, AI ain’t quite there yet. You just can’t rely on it to give you the right answers…until you get the hang of asking it the right questions in the right ways.
I’ve got a project to work on my bicycles. It’s a good time for it. Do some tweaking and setting up during the down times so everything will be ready for the next season. I’ll still be getting out on occasion, but more than likely I’ll really start moving in March.
I accidentally bought the wrong stuff to upgrade my Revolt — a GRX-812 rear derailleur…which turns out to be for 1x (which was not my intention) and the accompanying 11-42 rear cassette. Can’t use the derailleur (unless I convert the Revolt to a 1x, which was not my intention), so can’t use the cassette. But it turns out I do have a 1x — my old Trek (the bike before the Revolt). So why not set it up. ChatGPT tells me it’s entirely possible…until I tried to mount the 812 on the Trek, at which time I find out it needs a particular hanger that Trek absolutely does not make for that model and type of bike.
So the upgrade to the Trek is out too. And if I’d been more patient in asking ChatGPT (i.e., ask specifically “are you sure?” — in a few more words than that), then I probably wouldn’t have had my hopes up.
As it stands now, though, I still intend on working on the Trek — getting it back into good riding shape — and then working on the Revolt to give it a little more climbing ability. In the end, I’ll have two bikes going — an endurance bike for longer, tamer rides; and a gravel bike for some of the rougher. I look forward to the season, and especially working on the Trek, a bike on which I’m not afraid to experiment. That’s how it ended up 1x in the first place.
Just a year ago now, I was so proud of myself. I set up a Raspberry Pi as a cloud server and was able to access files on it from anywhere in the world. Way cool. Until I stopped maintaining it. Turns out that Pi’s are already notoriously hard on SD cards, and even if you don’t work them regularly…or it could be if you don’t work them regularly…they crap out on you. They may even look like they’re working (this morning I was able to connect with it) but when you reboot (as I did) they suddenly disappear.
Chester (ChatGPT) talked me through the whole thing. And I probably could’ve gotten it back up and running (limping)…until the next time it broke down..
So I asked Chet what I could do about it. I wanted to a system that was a little more robust, but still didn’t require the expertise of an IT guru. And it seems the solution isn’t too far off. At least my experience with the Pi has taught me the concepts of a cloud server and what needs to be happening, even if I don’t know all the terminology and bugs. And from what I understand, ChatGPT is getting even better and walking someone through the coding process than it was a year ago when I set the thing up the first time.
Now the question is, do I want a cloud server. I think I do. And for under a thousand bucks, I can set up a pretty good one. But instead of using a Pi, which is more of a hobbyist’s experimental product, I should be using a full-up mini PC (which doesn’t really on the SD card). And I think my experience with the Pi will help immensely when it comes to setting the new machine up. I know the questions to ask and I know what should be covered for security and reliability (firewalls, backups, certificates — that kind of stuff).
So again…do I really want the cloud? I mean, I had one for a portion of the past year and really didn’t use it enough to have it survive. Still, it’s a new year. I’ve got a nice MacBook on which I should be doing more things like blogging and writing and studying. It’s just that I’ve got to pay attention. Can I do it? Maybe. Should be a cool enough product, so…we’ll see.
