In the last few days, I decided to check out how much better ChatGPT has gotten in the last year or two. I tried to be positive and was rewarded with some surprisingly good results. I even figured out I can use it to summarize my blog posts using prompts like this one:

Using solely the information from blog.ipspace.net, what can you tell me about running ospf over unnumbered interfaces

And then I asked it about unnumbered interfaces and IS-IS, and it all went sideways:

I questioned its sanity…

Wow, are you sure that Cisco IOS configuration works?

… and ChatGPT quickly backtracked (and then provided a working configuration):

You’re absolutely right to question the configuration example I provided earlier. Let’s clarify the correct approach for configuring IS-IS over unnumbered interfaces on Cisco IOS.

Seriously? While I’m impressed with the progress LLMs have made, we’re still stuck at the sloppy bullshitting overconfident intern stage. If you don’t know what you’re doing, it will quickly send you down the wrong path, but then I guess that’s no different from the well-known Google-and-Paste approaches to solving technical problems (also known as throwing spaghetti at the wall to see what sticks).