Make It Routine

Question

Post reply

Make It Routine

Steve Jones - SSC Editor

SSC Guru

Points: 742868
More actions
July 27, 2026 at 12:00 am

#4884300

Comments posted to this topic are about the item Make It Routine

Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply

Site Owners SSC Guru Points: 80223 More actions · Answer 1

Thanks for posting your issue and hopefully someone will answer soon.

This is an automated bump to increase visibility of your question.

David.Poole SSC Guru Points: 76423 More actions · Answer 2

Part of the road to AI adoption is learning what you need to do to build trust in it and which bits you can trust. There are a lot of people out there claiming to be AI experts. I'm mindful of the time and effort it takes to become an expert, and I'm sceptical, especially when AI can produce some nasty surprises for people who ARE experts.

Some bits I'm still not comfortable with. The idea of allowing it to resize databases and remove logs is a prime example. Allowing it to build stuff is one thing. Allowing it to demolish something is a bridge too far at the moment.

There are some setup tasks I do once in a blue moon. I meant to write them down, but as blue moon tasks, I hadn't done so and always end up regretting it.

Without AI, these actions would probably have stayed in my list of eternal regrets, but with AI, what I've done is use it for the following:-

Draft the documentation, then worked through it, correcting any errors.
Build scripts to carry out basic setup tasks. Again, correct them where needed.

At this stage, I don't want AI to do everything. I want enough involvement, so I keep learning. A well-written script, commented to educate the person running it, is a good teaching tool.

LinkedIn Profile

Steve Jones - SSC Editor SSC Guru Points: 742868 More actions · Answer 3

resizing a db or removing/changing logs is interesting.

In general, we wouldn't want a human to do this directly, right? I'd want this run through a pipeline for a PR review and/or an approval. I'd think an AI could do the same thing. Submit changes into a process. An AI isn't really a computer process, it's a proxy for a human, so if we don't trust humans to do things without oversight (and we should less than we do), we shouldn't trust an LLM.