Relatedly if you build models, the datasets that you build to train them will outlast any model.

These days the life cycle of a model is sometimes minutes, or at the longest, months. Datasets last years, decades, or depending on the data, even centuries.

“Let me stop you right there…”

Caribbean nation Guyana is booming after discovering oil. BBC’s Stephen Sackur puts it to President President Dr Irfaan Ali; lobbyists say oil is bad for the climate.

Dude wasn’t having it. Mans was ready!

Reproducibility buffs: If you can't reliably do it again, did you really do it

Me admitting to my friend I only just started using Zotero to manage my paper readings yesterday:

“I been rawdogging academia until now it’s over for all of you”

ahem

VOTE TINO RANGATIRATANGA

What a great essay on colonialism in 21st century science

Yes, it’s three hours long but if you are surprised there’s that much to talk about you probably need to watch the video.

youtu.be/R7hK5_Rj--8?si…

Incidentally, this picture is a good illustration of my thoughts on whether mathematics is invented or discovered

For years I’ve been interviewing data annotation workers who are the lifeblood of the AI industry. For years I’ve heard the same story: the platforms they work for wield total power, leaving them precarious & vulnerable to exploitation. A horrible example of this just happened 1/

I've tried hard, but I've not come CLOSE to nailing the AI issue this well.

What do you mean you’re not schooled on 70s computing terminology?

(I liked the musicals question though I got that one ☺️)