Re: IPv8 / BGP8 / CF
Andrew Kirch via NANOG <[email protected]>
| Newsgroups | gmane.org.operators.nanog |
|---|---|
| Message-ID | <CALA5tJKJByw-G2CB0HmXFu0+WwXaa=UKxgsmPBmJ2L3aVZgfdA@mail.gmail.com> |
Andre's point here is 100% correct. This thread should probably die now. Andrew On Thu, Apr 30, 2026 at 7:21 PM Andre Sencioles via NANOG < [email protected]> wrote: > Dear Jamie, > > Stop talking to LLMs. No, seriously. Stop NOW! > LLM models WILL NOT tell you are wrong, they will keep reinforcing your > beliefs to keep you wasting tokens. This is what the models were "trained" > to do. > > You are showing all signs of going down a sycophantic loop with a > delusional LLM. You are not the first one, and you won't be the last. > I urge you to go and read this: > https://techcrunch.com/2025/10/02/ex-openai-researcher-dissects-one-of-chatgpts-delusional-spirals/ > But really ready it! Don't ask for a summary, or tell your agents to > extract the key points. Use your MK1 eyeballs. > > Now, after reading that carefully, think about this: > You have a whole community of experienced network engineers telling you > are wrong in the most fundamental ideas of your design. > > Ask yourself: Have I maybe got a bit too excited with my LLM sessions and > forgot to do some critical thinking? Am I so out of touch? Or is it the > children that are wrong? > > > With the best of intentions, > Andre > _______________________________________________ > NANOG mailing list > > https://lists.nanog.org/archives/list/[email protected]/message/TI4B6LGZUBKNR2JRJWD5J56VLIWU63QG/ > _______________________________________________ NANOG mailing list https://lists.nanog.org/archives/list/[email protected]/message/FSK3HBWI37T26LZL7WPL32OD2YVBM2XL/