On Sat, Apr 20, 2024 at 12:58:34PM +1000, Alexis wrote: > > and a bit surprinsigly - at least to me - chatgpt didn't get the syntax > > right either, no matter how detailed my prompt was. > > Not at all surprising to me, given that ChatGPT and other LLM-based 'AI' > systems - essentially Markov chains / glorified autocorrect - are > increasingly known for 'hallucinations' and confidently making false claims.
Here's the story of my asking it to write a PF.conf - https://nxdomain.no/~peter/chatgpt_writes_pf.conf.html or with nicer formatting and trackers https://bsdly.blogspot.com/2023/06/i-asked-chatgpt-to-write-pfconf-to-spec.html so in this context, near totally useless, likely due to insufficient volume of actually useful configurations in the data it was trained on. This other piece has it come up with some only tangentially related gibberish, but the thing partially redeems itself by offering up that poem at the end - https://nxdomain.no/~peter/chatgpt_on_ipv6_and_openbsd_poetry.html (or again with nicer formatting but G's trackers https://bsdly.blogspot.com/2023/03/chatgpt-opines-on-ipv6-procastination.html) - Peter -- Peter N. M. Hansteen, member of the first RFC 1149 implementation team https://bsdly.blogspot.com/ https://www.bsdly.net/ https://www.nuug.no/ "Remember to set the evil bit on all malicious network traffic" delilah spamd[29949]: 85.152.224.147: disconnected after 42673 seconds.