Rainer M Krug

@geospacedman@mastodon.social there is such a tendency here on Mastofon to shout BAAD as soon as AI is involved. To your question: it depends. I have seen so many bad slides generated by humans, that I would prefer good AI generated slides. But: the science behind them has to be good! Are we going to talks to be entertained by human made slides, or hear about good science?

@lwpembleton agreed. Yamls are great tool for that. Easy to read by humans and machine, easy brought into r and great for reproducibility and separating function from values / input.

@geospacedman just define unction <- function Problem solved.

@geospacedman@mastodon.social bad imagery and slides are bad. Good ones are good.

@brianmueller@mastodon.social great! But I would prefer Vinyl - a kickstarter?

@brianmueller@mastodon.social @CARROT@mastodon.social will it be in Apple Music? Can I download ( in two weeks time) the complete album? Will there be Vinyl? Or even on stage? To many questions, to many hopes!

@kirill@fosstodon.org works. Fixed a bug in duckdb (export to hive partitioned parquet dataset could lead to not readable dataset)

@geospacedman yup. Or loops. How often do you use ‘or’ in you code?

@addison @olibrendel you have a solid paper submitted and reviewer 2 wants more references . And there it starts.

@NanoRaptor@bitbang.social we need baby words - and the black sheep in the Words family : Libre Word

@Enzo90910@iosdev.space @softmaus@mastodon.social @neo@soc.psynet.me @gernot@mas.to are there any local coding agents, or can I run codex or Claude code with a local llm?

@creepy_owlet@mastodon.online @tomw@mastodon.social @geospacedman@mastodon.social I just saw the following which adds to the discussion. https://theoatmeal.com/comics/ai_art?no_popup=1

@LeoRJorge@mastodon.social @coatless@mastodon.social and llms are what can get us by out of this : why do I need shiny, if I can ask an llm to do the same using any other framework. Similar to tidyverse, but easier.

@tomw@mastodon.social @geospacedman@mastodon.social People can do good imagery (hand drawing, mermaid, plantuml, any gis for maps, etc) but are also some who can’t. There are different reasons why somebody uses AI for imagery - either they can’t (not their fault - if they have enough resources, they pay somebody), they do not have the time nor resources to pay anybody, they do not see them as essential in their presentation, etc. They also might not care. But it is not a “signifier” but rather indicative and not proven. So if you think the science is good, then you should be able to sit through it and listen (remember these overhead slides shows with text typed on them?).