@moozer

77 Followers
96 Following
544 Posts
Saying mooo in a modern way.
TopicsSecurity, networks, linux
Currently dabbling withansible, openbsd
Working withstudents and pros
Annual family trip#bornhack
Show threadmoozer5d ago
@falcennial let's hope a) doesn't become true, like if the AI companies (intentionaly) build in commercials and/or political propaganda.
Show threadmoozer5d ago
@JennyList I fear most people have not reached the peak yet,
moozerMay 15
Sune AukenMay 13

My research project, Writing With the Dead, has a fully funded PhD position available. Terribly exciting—and also with very nice people involved.

Come to Copenhagen. Do Grundtvig stuff. Wisen up.

#PhD #Grundtvig #ResearchIntegrity

Do repost, please.

https://jobportal.ku.dk/alle-opslag/?show=160458

PhD scholarship in Grundtvig Studies at the Faculty of Theology

moozerMay 12
kjaerulvMay 12
You know who you are....
moozerMay 12

I am struggling a bit with #githubactions. I want to be able to click "merge when checks pass", that requires that I specify which checks that must pass. These checks are by workflow.

I build and test multiple docker containers, and have a workflow for each. Making a workflow that includes everything is an option.

Must I make a master workflow?
there a trick to specifying check by name?

I get the sense that my thinking is wrong about how to structure github actions....

#github

Show threadmoozerMay 10

@tykling yes 😃

It is more a question if it is x10, x100 or (as I suspect) x1000 worse.

I found articles about how llm generated code compared to human written code, but nothing about abusing LLMs for tasks that "normal" scripts/programs could do.

moozerMay 10
Spring Jo 🥚  🍀May 9
My dad started quoting this and now I found the original
Show threadmoozerMay 9

@reynir jo, som du siger man en SMS og skal bruge den sammen med en kode.

Det løser mit problem med ikke at have #mitid app.

Mht. sikkerheden... bumbum... jeg stoler ikke videre på apps, og jeg kan forstå sms+kode, og jeg bruger unikke lange kodeord, så who knows.

moozerMay 9

I am looking for statistics or other resources that describe how much power or CPU cycles #AI and #LLM is using compared to a similar task performed by a regular script or program.

Anyone having seen such comparisons?

Show threadmoozerMay 8

@bert_hubert I'm starting to think in terms of process vs result. It seems to be well aligned with the author.

If only the end result is important (at some quality), AI could be a good tool. If the process and the associated learning is important, AI becomes potentially toxic and must be minimized.